GENERATIVE AI

IN PRACTICE
 GENERATIVE AI
IN PRACTICE
100+ AMAZING WAYS
GENERATIVE ARTIFICIAL
INTELLIGENCE IS CHANGING
BUSINESS AND SOCIETY

BERNARD MARR
This edition first published 2024
© 2024 by Bernard Marr
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted,
in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted
by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/
go/permissions.
The right of Bernard Marr to be identified as the author of this work has been asserted in accordance with law.
Registered Offices
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
For details of our global editorial offices, customer services, and more information about Wiley products visit us
at www.wiley.com.
Wiley also publishes its books in a variety of electronic formats and by print-­on-­demand. Some content that
appears in standard print versions of this book may not be available in other formats.
Trademarks: Wiley and the Wiley logo are trademarks or registered trademarks of John Wiley & Sons, Inc. and/
or its affiliates in the United States and other countries and may not be used without written permission. All
other trademarks are the property of their respective owners. John Wiley & Sons, Inc. is not associated with any
product or vendor mentioned in this book.
Limit of Liability/Disclaimer of Warranty
While the publisher and authors have used their best efforts in preparing this work, they make no
representations or warranties with respect to the accuracy or completeness of the contents of this work and
specifically disclaim all warranties, including without limitation any implied warranties of merchantability or
fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales
materials, or promotional statements for this work. This work is sold with the understanding that the publisher
is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable
for your situation. You should consult with a specialist where appropriate. The fact that an organization, website,
or product is referred to in this work as a citation and/or potential source of further information does not mean
that the publisher and authors endorse the information or services the organization, website, or product may
provide or recommendations it may make. Further, readers should be aware that websites listed in this work may
have changed or disappeared between when this work was written and when it is read. Neither the publisher
nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to
special, incidental, consequential, or other damages.
Library of Congress Cataloging-­in-­Publication Data:
Names: Marr, Bernard, author.
Title: Generative AI in practice : 100+ amazing ways generative artificial
intelligence is changing business and society / Bernard Marr.
Description: Hoboken, NJ : John Wiley & Sons, Inc., 2024.
Identifiers: LCCN 2024000295 (print) | LCCN 2024000296 (ebook) | ISBN
9781394245567 (cloth) | ISBN 9781394254255 (adobe pdf) | ISBN
9781394254248 (epub)
Subjects: LCSH: Artificial intelligence—Economic aspects. | Artificial
intelligence—Social aspects.
Classification: LCC HC79.I55 M36528 2024 (print) | LCC HC79.I55 (ebook) |
DDC 658/.0563—dc23/eng/20240111
LC record available at https://lccn.loc.gov/2024000295
LC ebook record available at https://lccn.loc.gov/2024000296
Cover Design: Wiley
Cover Image: © Jannatul Koraise/Adobe Stock
Set in 11/14 pts Minion Pro by Straive, Chennai, India
 To my wife, Claire, my anchor in the shifting seas of life, whose love lights
the way through every storm. And to my children, Sophia, James, and
Oliver, my brightest stars in the vast universe, who remind me every day that
the most profound truths often come from the simplest curiosities.
 CONTENTS

About the Author ix

Introductionxi

Part 1 Setting the Scene for an AI Revolution1

1 Unveiling Generative AI: A New Frontier 3
2 Tracing the Evolutionary Blueprint of Generative AI 17
3 Revolutionizing Societies and Business Ecosystems 29
4 Risks and Challenges to Manage 45
5 Impact of Generative AI on Jobs 63

Part 2 Generative AI in Practice 77

6 A New Dawn in Media and Entertainment 79
7 Advertising and Marketing: Bridging Creativity with AI 97
8 Reinventing Customer Engagement through Intelligent Systems 109
9 Retail Reimagined: The Rise of Virtual Try-­ons, AI Shopping
­Assistants, and More 121
10 Personalized Learning: The Future of Education 133
11  Healthcare Transformation: From Personalized Advice to
Operational Improvements 143
12 Video Game Design and Testing: The Generative AI Approach 155
13 The Legal Sphere: AI-­Assisted Document Creation and Review 165
14 Crafting Tomorrow: AI in Design and Development 173
15 Banking and Financial Services: AI as a Disruptive Force 189
16 Coding and Programming: The AI Revolution 201
17 Data Insights: Harnessing the Power of Generative AI 211

vii
 Contents

Part 3 Moving Forward with Generative AI223

18 Implementing Generative AI: Keys for Success 225
19 
Glimpses of the Future: Predicting the Trajectory of
Generative AI 241

Appendix: GenAI Tools 249

Index271

viii
 ABOUT THE AUTHOR

Bernard Marr is a world-­renowned futurist, influencer, and thought leader

in the fields of business and technology, with a passion for using technol-
ogy for the good of humanity. He is a multi-­award-­winning and internation-
ally bestselling author of over 20 books, writes a regular column for Forbes,
and advises and works with many of the world’s best-­known organizations.
­Bernard has a combined subscriber base of 4 million people across his social
media channels and newsletters and was ranked by LinkedIn as one of the top
five business influencers in the world.

Bernard helps organizations and their management teams prepare for future
trends and create the strategies to succeed. He has worked with or advised
many of the world’s best-­known organizations, including Amazon, Microsoft,
Google, Dell, IBM, Walmart, Shell, Cisco, HSBC, Toyota, Nokia, Vodafone,
T-­Mobile, the NHS, Walgreens Boots Alliance, the Home Office, the Ministry
of Defence, NATO, and the United Nations, among many others.

Connect with Bernard on LinkedIn, X (@bernardmarr), Facebook, ­Instagram,

and YouTube to take part in an ongoing conversation; subscribe to Bernard’s
podcast; and head to www.bernardmarr.com for more information and hun-
dreds of free articles, whitepapers, and e-­books.

If you would like to talk to Bernard about any advisory work, speaking
engagements, or influencer services, please contact him via email at hello@
bernardmarr.com.

ix
 INTRODUCTION

It feels like generative AI has come out of nowhere, doesn’t it? Perhaps you
were happily going through life thinking artificial intelligence was something
other people had to worry about. And then boom! Suddenly, everyone’s talk-
ing about this new thing called “ChatGPT” and how it’s going to take over
everyone’s jobs.

As you’ve no doubt noticed, there’s a lot of hype around generative AI (or

GenAI as I call it in this book), and tools like ChatGPT. I expect you’ve picked
up this book because you want to cut through the chatter and understand just
what the deal is with GenAI. Is it really worthy of the hype? How might GenAI
impact your job? Should your organization be leveraging this technology?

To answer the first question, yes, GenAI really is worthy of all the noise.
I believe it’s one of the most powerful, transformative technologies that
humans have ever had access to. It’s right up there with the emergence of the
internet. Seriously! We’ll cover the second and third questions throughout
this book but, spoiler alert, GenAI (probably) won’t take your job. That said,
it will almost certainly change the way you work. And yes, your organization
should absolutely be looking to harness this transformative technology. That’s
why I wrote this book.

The Incredible Impact of Generative AI

I’m not the only one convinced that GenAI is utterly transformative.

One study led by Harvard found that GenAI tools helped workers at the
Boston Consulting Group gain a 40% performance boost.1 We’ll see many

xi
 INTRODUCTION

examples throughout the book of organizations using GenAI to accomplish

work more easily, lighten the burden of mundane tasks, and provide a better
employee (and customer) experience.

GenAI will be so ubiquitous that Bloomberg predicts the GenAI market will
explode from £40 billion in 2022, to a whopping US$1.3 trillion by 2032.2
What’s more, according to research by McKinsey, GenAI could add up to
US$4.4 trillion in value to the global economy annually – and save up to 70%
of workers’ time.3 It’s no wonder McKinsey refers to GenAI as “the next pro-
ductivity frontier.”

That’s the economic impact, but what about our everyday lives and the way
we do business? That, too, will be fundamentally altered by GenAI. We’ll talk
about the business and societal impact in Chapter 3, but to whet your appe-
tite, GenAI will change how we work, how we shop, how we consume content,
how we experience healthcare, how we learn (at school and otherwise), how
we play video games, and even potentially how we date. Adam Selipsky, CEO
of Amazon Web Services, says GenAI will “pretty much change every appli-
cation inside of companies, every application that consumers interact with. It
is a profound technological change that we’re still at the very beginning of.”4
I couldn’t agree more.

But wait, back up a minute . . . US$4.4 trillion in economic benefit? I know

what you’re thinking. How the heck can chatbots add so much value?

Because Generative AI Is More than Chatbots

We’ll delve into a more detailed definition of GenAI in Chapter 1, but for
now, all you need to know is that GenAI is a facet of artificial intelligence that
allows computers to create content, including (but not limited to) text. One of
the ways in which this technology can be harnessed is through more intuitive,
more responsive chatbots – chatbots with advanced language capabilities that
can respond to complex queries, and understand context as well as content.

ChatGPT is a chatbot. Developed by OpenAI and available free for public

use (although a paid-­for, upgraded version is available), it’s probably the most

xii
 INTRODUCTION

advanced chatbot we’ve ever seen. It can answer almost any question you throw
at it. It’s brilliant. It’s a great example of GenAI in action. However, GenAI is
much bigger than ChatGPT. And its uses extend far beyond chatbots.

Therefore, this is not a book about ChatGPT (although ChatGPT will obvi-
ously feature in several examples throughout). This book shines a light on the
vast array of tasks that GenAI tools can accomplish – from designing video
games to interpreting medical scans, and more. Yes, GenAI gives computers
the ability to converse and answer questions. But it can do so much more on
top of that.

Almost every time I type “GenAI,” I think of the word “genie.” And for good
reason. Working with GenAI is a bit like having your own personal genie to
help lighten the load, get more done, and generally make life easier. (If only it
could grant infinite wishes. . .)

That said, as we’ll see in this book, getting the best out of GenAI requires a
balance between human expertise and machine intelligence. This isn’t about
total automation – it’s about humans collaborating with GenAI to do things
better. At the end of the day, GenAI is just a tool. An incredible tool. But a tool
nonetheless – and it’s up to us to use it thoughtfully.

With that in mind, my hope is that GenAI will be used not just as a productiv-
ity tool, but as a force for good in our world. As an example, in Chapter 11,
we’ll see how GenAI can be used to democratize access to healthcare for peo-
ple around the world. And in Chapter 14, we’ll see how GenAI is speeding up
the process of discovering life-­saving new drugs. True, it can also be used to
generate fake content to sway elections or be harnessed by criminals for all
sorts of nefarious purposes. But GenAI also has the potential to help solve
some of humanity’s biggest challenges.

What to Expect from This Book

The bulk of this book focuses on practical applications of GenAI across a
wide range of sectors. In other words, we’ll explore how organizations are
already using this technology – and, in the process, help you unearth

xiii
 INTRODUCTION

potential use cases for your own organization. That’s the focus of Part 2 of
this book, ­practical examples and takeaways from other organizations that
have deployed GenAI.

Then in Part 3, we’ll look at keys for success when implementing GenAI in
your own business. Plus, you’ll find predictions for the future of GenAI.

But before all that, Part 1 sets the scene for this powerful, transformative tech-
nology. We’ll explore what the technology is, how it works, how it will change
our world (and the world of business), and how it will impact jobs. We’ll also
take an honest look at the risks and challenges that surround GenAI.

Before you tuck into Part 1, it’s important to note that GenAI technology is
rapidly evolving. And it’s getting better and better all the time. Indeed, one
of the things that made this book challenging to write is the fact that the
technology is advancing so darn quickly. Amazing new use cases and updates
to GenAI tools were emerging literally every week. As an example, I’d barely
finished writing Part 1 when OpenAI launched a new version of ChatGPT
that could not just answer written questions, but also understand speech and
visuals, and respond in an artificial voice.

Safe to say, by the time you read this, new tools and new examples will have
surfaced that continue to surprise and delight. It’s truly an exciting time.

Notes
1. Enterprise workers gain 40 percent performance boost from GPT-­4, Harvard
study finds; Venture Beat; https://venturebeat.com/ai/enterprise-­workers-­gain-­
40-­percent-­performance-­boost-­from-­gpt-­4-­harvard-­study-­finds/
2. Generative AI to become a $1.3 trillion market by 2032, Research finds; B ­ loomberg;
https://www.bloomberg.com/company/press/generative-­ai-­to-­become-­a-­1-­3-­
trillion-­market-­by-­2032-­research-­finds/
3. The economic potential of generative AI: The next productivity frontier; M ­ cKinsey;
https://www.mckinsey.com/capabilities/mckinsey-­digital/our-­insights/the-­
economic-­potential-­of-­generative-­ai-­the-­next-­productivity-­frontier
4. AWS CEO: Generative AI will ‘pretty much change’ every application consumers
interact with; YouTube; https://www.youtube.com/watch?v=jQ18HJkLJjw

xiv
 Part 1
SETTING THE SCENE
FOR AN AI REVOLUTION

I said in my introduction that Generative AI (GenAI) is one of the most

­powerful technologies humans have ever had access to. But what exactly is
GenAI and how does it work? How did we get to this point? And how, broadly
speaking, might GenAI change our world?

In Part 1, we’ll answer all those questions and more. We’ll define GenAI and
how it works, explore the evolution of GenAI until this point, outline the risks
associated with GenAI, and take a big-­picture look at how GenAI is going
to change our world – including the jobs that humans do. This will set the
scene nicely before we delve into GenAI use cases across various sectors and
industries in Part 2.

1
 1
UNVEILING GENERATIVE
AI: A NEW FRONTIER

Okay, let’s go back to basics. What is generative AI (artificial intelligence)?

How does it work? And what is the technology capable of? You’ll find out
in this chapter as we delve under the hood of GenAI, tug at a few wires, and
examine what the heck’s going on in there.

I really want this chapter to give you an inspiring feel for the many things that
GenAI is capable of – but also to drive home the point that GenAI isn’t just
about ChatGPT (Chat Generative Pre-­trained Transformer). Sure, ChatGPT
is a prime example of GenAI (and it certainly hoovers up the majority of
GenAI’s press), but there’s a lot more to GenAI than ChatGPT . . . as you’ll
find out in this chapter.

What Is Generative AI? A Quick Explanation

Actually, let’s start by defining artificial intelligence (AI) in its ­broadest
sense. The term “AI” refers to computer algorithms that can effectively
simulate human cognitive processes, like learning, decision-­making, and
problem-­solving.

GenAI is a groundbreaking subset of AI – the cutting edge of the cutting

edge – that is able to create new content based on patterns and structures it
has learned from existing data. Like any AI, GenAI tools are given enormous
amounts of data to learn from (what’s known as “training data”). They learn

3
 GENERATIVE AI IN PRACTICE

from the training data, and then use the patterns or rules that they’ve learned
to create new content that’s similar to, but not exactly the same as, the data
they have been trained on.

An example or two
A good example is DALL-­E 2 – the text-­to-­art platform that allows anyone to
generate artworks. Or, of course, there’s ChatGPT, the language model that
can create text based on conversational text prompts. (You’ll find a much big-
ger list of GenAI tools in the Appendix, by the way.) These systems learn
from huge training datasets – ChatGPT, for example, was trained on vast
amounts of text from the internet, including web pages, articles, and books.

Text and images are perhaps two of the best-­known uses of GenAI so far, but
the technology is capable of so much more. With GenAI you can generate
product designs, computer code, music, video, voices, and even entire visual
worlds. We’ll talk more about GenAI’s capabilities later in the chapter, but to
whet your appetite, imagine being able to create unique video game worlds
rendered in real time, or have a book written just for you, or have your favorite
celebrity read you today’s news. The possibilities are mind-­blowing.

The capabilities are already quite impressive, but in the future, GenAI systems
will be able to create pretty much anything that humans can. And this, in
turn, means GenAI can turn anyone and everyone into an author, musician,
computer programmer, filmmaker, or other type of creator.

How GenAI differs from the AI we are used to

AI is increasingly part of the world around us, including the search results
you get on your phone, your conversations with Alexa, and the movie
­recommendations that Netflix serves up on an evening. So what makes
GenAI different to these “traditional” AI tools? (I realize it sounds strange to
refer to “traditional AI” when it’s hardly been around that long, but I do so
to distinguish between the AI that we’re already used to in everyday life and
this new evolution of AI systems. One technical term for traditional AI is
“discriminative AI.”)

4
 UNVEILING GENERATIVE AI: A NEW FRONTIER

Traditional AI systems also learn from large amounts of data, but they deliver
a different output. Traditional AI systems are used to make predictions based
on existing data. And we use those predictions to help us make better deci-
sions, at work and in everyday life. This could span anything from listening
to new music on Spotify, and viewing recommended products on Amazon,
to identifying which of your company’s customers are most likely to buy a
certain product.

This new wave of GenAI tools goes even further, by creating new content
based on existing data. In other words, GenAI isn’t just about simulating
human cognitive processes like decision-­making and problem-­solving – it’s
about simulating human creativity.

To further illustrate the difference, imagine you’re playing computer chess.

The computer knows all the rules, can predict your moves, and makes its own
moves. It’s not inventing new chess moves; rather, it’s selecting the right move
based on existing strategies. That’s traditional AI – it’s like a master strategist
who can make smart decisions within a set of rules. And it does it very well.
But GenAI? Well, that could, in theory, come up with new ways of playing
chess that we haven’t invented yet.

So, traditional AI excels at pattern recognition, while GenAI excels at pattern

creation. Which is very cool indeed.

That said, GenAI and traditional AI aren’t mutually exclusive. They could
be used in tandem to provide even more powerful solutions. For example, a
traditional AI could analyze user behavior from your company’s website, and
then a GenAI could use the analysis to create personalized content for users.

How Does Generative AI Work?

Think of it like learning to draw by looking at many pictures. After looking
at many pictures, you try to draw something new on your own. GenAI does
something similar: it “looks” at a lot of data (text, visual, or other), learns the
patterns within that data, and then tries to create something new that fits
those patterns.

5
 GENERATIVE AI IN PRACTICE

So, in very simple terms, GenAI is like an artist or writer that has studied lots
of existing works and then tries to create its own work based on what it has
learned. This process is powered by complex algorithms that mimic how our
brain works in order to learn from data and identify patterns.

That’s the super-­simple explanation. Let’s get a little more technical.

Introducing machine learning and neural networks

We’ll talk more about the evolution of GenAI in Chapter 2, but as a quick
primer, GenAI grew out of a field of AI study and practice called “machine
learning” – indeed all of the AI we see today is based on machine learn-
ing. While traditional computer algorithms are coded by a human to tell a
machine exactly how to do a particular job, machine-­learning algorithms are
able to make decisions based on what they learn from the data. The more data
they’re fed, the better they get at this process.

Another term you’ll need to get to grips with is “neural networks,” as this is
the core technology that GenAI is built upon. A neural network is essentially
an advanced machine-­learning model inspired by the way human brains
work. While a less complex machine-­learning model may need some human
intervention in the process, a neural network has the ability to learn and make
decisions by itself, and can even learn from its own errors – rather like the way
a human learns through a process of trial and error.

Here’s how a neural network works:

• Training: neural networks learn through a process called “training.”

During training, the neural network is fed a lot of data (which could
be text, images, or whatever), and it learns to identify patterns and rela-
tionships in the data.
• Learning: as the neural network is exposed to more and more data –
and we really are talking about vast amounts of data – it gradually gets
better at identifying patterns and understanding the underlying rules
that govern the data.

6
 UNVEILING GENERATIVE AI: A NEW FRONTIER

• Layers: neural networks are organized into layers, and each layer is
responsible for identifying different types of patterns. The initial layers
might identify simple patterns, and as we move deeper, the layers are
able to identify more complex patterns.
• Generative models: GenAI often uses specific types of neural networks
called “generative models.” In addition to recognizing patterns, gen-
erative models are able to generate new data that is similar to (but not
exactly the same as) the data they were trained on.
• Input and output: once the generative model is trained, you can give
it an input (e.g., a partial image or a text prompt), and it will generate
an output (like a completed image or a piece of text) based on what it
learned during training.
• Randomness: GenAI introduces a certain amount of randomness in
the generation process, which means that it can produce slightly dif-
ferent outputs each time, even when given the same input over and
over again.

Examples of generative models

I just mentioned generative models, the neural networks that enable GenAI
to create new content. Here are some examples of generative models used in
GenAI applications:

• Large language models (LLMs): by gobbling up large amounts of text,

LLMs learn the semantic relationships between words and use that data
to generate more language. An example of an LLM is GPT-­4, created by
OpenAI, which powers the ChatGPT tool.
• Generative adversarial networks (GANs): these work by pitting two
competing algorithms against each other, one tasked with generating
data that resembles its training data and the other tasked with trying to
tell whether the output is real or generated. This type of model is typi-
cally used to create images, sounds, or even video (including deepfakes).
• Variational autoencoders: this is a type of model that learns how data is
constructed. It’s often used to generate synthetic data.

7
 GENERATIVE AI IN PRACTICE

• Diffusion models: these work by adding random data (known as

“noise”) to the data it’s learning about, and then figuring out how to
remove it while preserving the original data. This way, the model learns
what’s important and what can be discarded. Diffusion models are most
commonly used in image generation.
• Foundation models: this is something of an umbrella term for m ­ odels
that are pre-­trained on a broad range of data at an enormous scale
and then fine-­tuned for specific tasks. These are large-­scale, general-­
purpose models that are the foundation of LLMs and other GenAI tools.

With different models able to do different things, it’s clear that GenAI has a
wide range of applications. Which brings us to. . .

What Can Generative AI Do?

In Part 2, we’ll delve into how GenAI is already being used in practice (with
lots of real-­world examples), but for now, I really want to make the point that
GenAI is capable of creating all sorts of content.

GenAI is so much bigger than ChatGPT

Of course, ChatGPT gets a lot of attention – and for good reason: it’s an
impressive tool, and creating text is a major application of GenAI. Which is
probably why “ChatGPT” and “generative AI” have almost become synony-
mous terms. But GenAI is capable of so much more than writing.

Take Coca-­Cola’s Masterpiece advert, for example. A collaboration between

human artists and GenAI, the ad brings many of history’s greatest artworks to
life on the screen in a way that’s never been done before. GenAI has also been
used to create a new Beatles song by rebuilding partially recorded lyrics by
John Lennon, combined with new material by Paul McCartney.

But GenAI isn’t just relevant to creative content. It’s also being used to speed
up drug discovery, with one UK company recently announcing that it’s c­ reated
the world’s first AI-­generated immunotherapy cancer treatment.

8
 UNVEILING GENERATIVE AI: A NEW FRONTIER

Then there’s “generative design” – an emerging field where GenAI is used to

create designs and production processes for new products. In one instance,
General Motors used generative tools created by Autodesk to design a new seat-
belt bracket that’s 40% lighter and 20% stronger than its existing components.

And we can’t talk about GenAI without talking about “deepfakes,” which blur
the lines between reality and fiction by making it appear as if real people have
done or said fake things. GenAI is the technology behind deepfakes, and it’s
only going to get better at creating fake – but incredibly realistic – ­content.
Deepfake Tom Cruise was one of the earliest and most famous examples
(Google it if you haven’t seen it – it’s uncanny). More insidiously, potential
candidates on both sides of the 2024 US presidential elections have starred in
deepfakes designed to discredit them.

As if spreading false information wasn’t bad enough, there are also outright
criminal applications of GenAI, including attempts to extort money by stag-
ing hoax kidnappings using cloned voices, and, in one memorable example,
fraudulently scamming money by posing as a company CEO.

As the technology has become more sophisticated, GenAI’s capabilities have

expanded greatly. To summarize, GenAI can be used to generate:

• Text
• Images
• Video
• Music and sound
• Designs
• Data analysis
• Computer code
• Synthetic data
• Research
• Video games and virtual worlds

Let’s explore each usage in a little more detail.

9
 GENERATIVE AI IN PRACTICE

Text – writing text that’s (pretty much) indistinguishable

from a human writer
Alongside ChatGPT, there are other generative text tools like Google’s Bard
and Meta’s Llama. These tools can be used to write anything from essays and
articles, to plays, poems, and novels.

Here are some of the incredible things that generative text tools can do for you:

• Writing assistants: helping you draft emails, reports, and other written
materials efficiently.
• Creative writing: assisting authors in coming up with unique story
ideas, characters, plotlines, or even drafting entire narratives.
• World-­building: aiding authors in creating vivid and immersive worlds
for their narratives, with tools that help outline histories, geographies,
and cultures of fictional settings.
• Writing prompts: offering a variety of writing prompts to help writers
overcome creative blocks and generate new ideas.
• Grammar and style corrections: offering real-­time suggestions for improv-
ing the grammatical accuracy and stylistic fluency of written content.
• Translation assistance: help with translating texts to different languages
while retaining the nuances of the original content.
• Accessibility: providing features like speech-­to-­text to help people with
disabilities draft texts.
• Song writing: creating song lyrics, even in the style of your favorite artist.
• Recipe writing: coming up with recipes that suit your parameters (e.g.,
“a dinner recipe that uses butternut squash, feeds a family of four, and
takes no longer than 45 minutes”).
• Plain-­English explanations and text summaries: helping you under-
stand a complex topic in plain English, or providing a succinct sum-
mary of a long piece of text.
• Multiple-­ choice questions: generating quiz or test questions for a
­variety of topics, to beginner, intermediate, or advanced levels.

10
 UNVEILING GENERATIVE AI: A NEW FRONTIER

• Travel itineraries: aiding you with everything from finding flights and
accommodations to suggesting popular attractions to visit. Even where
to get a good meal, based on your dietary needs and budget.

Images – creating images to your specifications

Many GenAI tools – such as Midjourney or Stable Diffusion – can take a
natural language prompt (i.e., regular, conversational language) and use it to
generate a picture. Tell it you want an image of a two-­headed dog wearing an
Elvis costume flying a spaceship into a black hole and watch as it (or some-
thing close to it) appears before your eyes.

This can be useful for:

• Art creation: assisting artists in generating unique artworks by under-

standing and mimicking different art styles.
• Style transfer: enabling artists to apply the styles of famous artworks to
their creations.
• Automated sketching: assisting in automatically generating sketches
based on the descriptions provided, aiding artists in the initial stages
of their work.

Video – from editing to creation

While not yet as advanced as text or image generation, tools are beginning to
emerge that allow us to create and edit video simply by describing what we
want to see.

Typical uses include:

• Enhanced editing: automating intricate editing tasks including cut

detection, color correction, and subtitle generation, thereby streamlin-
ing the post-­production process.
• Dynamic special effects: allowing real-­time rendering of high-­quality
special effects, motion tracking, and 3D modeling, significantly reduc-
ing the time and effort involved in post-­production.

11
 GENERATIVE AI IN PRACTICE

• 3D modeling: facilitating the creation of 3D models and environ-

ments for integration into videos, helping to create immersive visual
experiences.
• Sound design and synthetic voice: enhancing audio experiences
through AI-­generated music compositions and synthetic voices. Which
brings us to. . .

Music and sound – from voices to songs

GenAI tools can create human-­like voices (voice synthesis), allowing com-
puters to speak words. They can also create music and sound effects.

In practice, this can be used for:

• Harmony suggestions: proposing harmonious chord progressions and

melody lines.
• Generative music: creating entire musical pieces automatically based
on a set of input parameters.
• Mood-­based creation: offering tools that help composers create music
tailored to specific moods or emotions.
• Voice synthesis: generating realistic voices that can read text naturally,
providing a human-­like auditory experience.
• Language and accent variability: offering a wide range of language
and accent options, helping to create diverse and inclusive auditory
experiences.
• Customizable voices: providing tools to create customizable voices,
allowing users to personalize their auditory experiences.

Graphic design and generative design – streamlining

design work
GenAI offers huge potential for graphic designers and product designers to
streamline the design process – from automating aspects of visual design to
creating multiple product design variations.

12
 UNVEILING GENERATIVE AI: A NEW FRONTIER

Some of the ways this may prove useful include:

• Design suggestions: aiding designers by providing design suggestions.

• Color palette suggestions: providing suggestions for color palettes that
work well.
• Stock image integration: offering seamless integration with stock image
libraries, giving designers easy access to a wide range of resources.
• Architectural design: assisting architects in generating building designs
optimized for specific criteria like materials cost, energy efficiency, and
spatial requirements.
• Product design: helping designers to come up with optimized and
innovative product designs by exploring a vast number of design alter-
natives in a short time.
• Fashion design: assisting fashion designers by proposing new clothing
designs based on current trends, historical data, and designer inputs.

Data analysis and reporting – democratizing data analysis

With GenAI tools, you don’t need to be a data scientist to interrogate data.
From generating automated reports to providing useful predictions, there are
lots of ways GenAI can help businesses get the most out of data.

In terms of automated reporting, GenAI offers:

• Real-­time reporting: offering tools that generate reports in real-­time,

providing up-­to-­date information for decision-­making.
• Visual representation: creating visual representations of data, such as
graphs and charts, to make reports more understandable and engaging.
• Natural language summaries: generating summaries of reports in easy-­
to-­understand language.

And in terms of predictive analysis, GenAI can deliver:

• Anomaly detection: identifying unusual patterns in data that may indi-

cate potential issues or opportunities.

13
 GENERATIVE AI IN PRACTICE

• Forecasting and customer insights: utilizing historical data to make

­accurate forecasts about future trends, and identify customer preferences.
• Simulation and modeling: creating detailed simulations based on data
analysis, allowing organizations to explore potential scenarios and
make better decisions.

Coding – computer programming made simpler

As well as ChatGPT, tools like Microsoft’s GitHub Copilot and Amazon’s
CodeWhisperer make it easy for anyone to generate computer code with very
little technical knowledge.

This can span:

• Code generation: creating code based on high-­level descriptions or

partially completed code.
• Code review: assisting in reviewing codes by identifying bugs, security
vulnerabilities, and providing suggestions for optimization.
• Automated testing: creating tests automatically based on the software’s
specifications, helping to improve the software’s reliability.

Synthetic data – solving privacy issues and more

GenAI makes it easy to create entirely synthetic data for use in training other
AI models. And for everyday businesses, synthetic data can help them over-
come the privacy and data security obligations that come with real-­world data.

In practice, this may include:

• Data augmentation: creating additional data through techniques like

image augmentation to enhance the training of AI models.
• Privacy preservation: generating synthetic data that preserves the pri-
vacy of individuals by not using real personal data.
• Balanced datasets: creating balanced datasets that help in reducing
biases in AI models, promoting more fair and equitable AI systems.

14
 UNVEILING GENERATIVE AI: A NEW FRONTIER

• Diverse data: generating diverse datasets that help researchers explore a

wide range of scenarios and conditions.

Research – making discovery easier and faster

There are many ways GenAI can aid the research process, with a few notable
examples being:

• Drug discovery: speeding up the drug discovery process by predicting

potential drug candidates through the analysis of vast datasets.
• Simulation environments: creating realistic simulation environments based
on synthetic data, aiding researchers in conducting experiments and studies.
• Rare event modeling: assisting in modeling rare events through syn-
thetic data, helping researchers explore scenarios that are difficult to
study with real data.

Video games and virtual world generation – making

immersive content more responsive
From virtual reality (VR) environments and immersive video game worlds
to the wider concept of the metaverse, we’re increasingly interacting with
immersive, highly realistic content. Designing such content is a complex task
that can be accelerated with the help of GenAI. Here’s how:

• Procedural generation and design: creating intricate environments,

realistic terrains, and adaptive storylines, fostering unique and immer-
sive gameplay experiences.
• Character design and animation: generating natural character anima-
tions, and crafting consistent character personalities to enhance story-
telling and immersion.
• Personalized user experiences: tailoring games and immersive environ-
ments to individual preferences and behaviors, fostering a more per-
sonalized and engaging user journey.
• Automated content moderation and security: monitoring and main-
taining the safety of immersive environments, using automated systems
to detect and manage inappropriate content.

15
 GENERATIVE AI IN PRACTICE

The Impact of Generative AI

GenAI is clearly capable of amazing things and has the potential to transform
business and everyday life. I’ll explore examples from specific sectors across
Part 2, but for a broader overview, turn to Chapter 3.

The rapid evolution of GenAI is also forcing us to confront some difficult

questions, and fast. For example, what happens when we reach a point where
it’s impossible to tell the difference between what’s real content and what is
generated by AI? There are no easy answers right now, but we’ll do our best to
delve into such thorny issues in Chapter 4.

And then there’s the question of how all this will affect human jobs, as
machines take on increasingly creative work. That’s the subject of Chapter 5.

Key Takeaways
In this chapter, we’ve learned:

• GenAI is a groundbreaking type of AI that is able to create new content

based on patterns it has learned from existing data.
• While traditional AI simulates human cognitive processes like decision-­
making, GenAI simulates human creativity.
• GenAI is made possible by neural networks – an advanced machine-­
learning model inspired by the way human brains work.
• ChatGPT may have made GenAI famous, but GenAI is much bigger
than ChatGPT. GenAI can be used to create music, sound, and video,
write computer code, design products, create immersive worlds, auto-
mate data analysis and reporting, and more.

Before we move on to the impact of GenAI and the issues surround-

ing it, let’s briefly explore the evolution of GenAI, from the early days of
­computing to now.

16
 2
TRACING
THE EVOLUTIONARY
BLUEPRINT
OF GENERATIVE AI

We’ve reached the point where AI is no longer just an executor of tasks or a

decision-­making tool, but also an inventor. This is an age where AI can be
a creative companion, capable of producing original outputs that can inspire,
assist, and even amaze us. In the next chapter, we’ll look at how this devel-
opment will shape our world. But to navigate the future of GenAI, we need
to understand its past. How did we get to this point? How far back do the
roots of GenAI stretch? (You might be surprised to discover just how far.)
And how is GenAI’s evolution inextricably linked with other fast-­developing
technologies?

Come with me as I take you on a tour through the history of GenAI, and
explore the key milestones that have brought us to this point.

1950s–1990s: The Early Years of AI

Scientists and researchers first introduced the concept of AI and machine
learning in the 1950s, which goes to show that humans have been fascinated
by the concept of intelligent machines for a long, long time.

17
 GENERATIVE AI IN PRACTICE

The promise of AI takes root

British mathematician, logician, and computer scientist Alan Turing devised
the “Turing Test” in 1950 – a method for determining whether a computer
can demonstrate human-­like intelligence. The test goes like this: a human
evaluator engages in a natural language dialogue with an unseen partner –
which could be a human or a computer – and if the evaluator cannot reliably
tell the machine from the human, the machine is said to have passed the test.
The Turing Test wouldn’t be passed for decades, reportedly by an AI in 2014
(although experts debate whether that particular AI really did pass the test).
But with the incredibly advanced tools available today – tools like ChatGPT –
I think we can safely say we’ve surpassed the test, at last.

Then in 1956, the Dartmouth Conference gathered around 100 bright minds
together to foster the idea that machines could mimic human intelligence.
This event founded AI as a distinct academic discipline in its own right.

Building on that, the 1960s and 1970s saw the development of early AI
­programs ELIZA and SHRDLU. Both were examples of natural language
­processing – indeed, ELIZA can be considered the world’s first primitive
­chatbot – but neither were sophisticated enough to pass the Turing Test.

The rise of expert systems

Then the 1970s and 1980s brought increasing computing power and the avail-
ability of personal computers, thus democratizing access to technology and
setting the stage for widespread research and development in AI.

The AI models developed in the 1970s and 1980s drew upon human exper-
tise in order to make decisions in specific areas, such as medical diagnosis or
credit scoring. But, for these expert systems to work, they had to be coded with
all the answers (e.g., all the symptoms associated with a disease), rather than
learning for themselves. AI still had a long way to go, in other words. So it’s
no great surprise that, as a result of unmet expectations and technical limita-
tions, funding and interest in AI research dropped off at times during these
decades – most notably in the mid-­1970s and again in the late-­1980s. These
periods are known as “AI winters.”

18
 TRACING THE EVOLUTIONARY BLUEPRINT OF GENERATIVE AI

The dotcom revolution

Then in the 1990s we saw the advent of the World Wide Web (WWW).
As well as ushering in a new era of connectedness (and cat memes) the inter-
net played a pivotal role in the evolution of AI. Why? Because the internet cre-
ated an exponentially growing pool of data, which became fertile ground for
training machine-­learning models. This drove advancements in data mining
and predictive analytics in the 1990s – with algorithms being programmed to
analyze large datasets, identify patterns, and make predictions. This would lay
the groundwork for the AI revolutions to come, allowing machines to learn
from data without being taught absolutely everything by a human.

2000s–2010s: Laying the Foundations

for Generative AI
So far we’ve explored the developments that allowed AI as a discipline to
flourish. Now we get to the developments that paved the way for the sophisti-
cated GenAI systems we see today.

Big data goes . . . big

In the 2000s, as more and more of our everyday activities moved online, the
resulting explosion in big data provided a huge boost for the evolution of AI.
As we know, more data means more information for training and fine-­tuning
AI algorithms, which in turn enabled algorithms to analyze complex pat-
terns and offer more accurate predictions and insights. This allowed machine
learning to find more practical applications in a wide array of fields, ­including
healthcare, finance, and ecommerce (think Amazon’s personalized recom-
mendations, and the like).

Deep learning marks a turning point

This explosion in data also enabled the advancement of deep learning in the
2010s. Deep learning is a subset of machine learning, which enables comput-
ers to learn from a large amount of unlabeled data. So, instead of a human

19
 GENERATIVE AI IN PRACTICE

telling the computer everything it needed to learn, the creation of deep neural
networks – which mimic the way the human brain works (see Chapter 1) –
allowed machines to interpret and analyze complex patterns in large data-
sets, without being told what to look for. This meant machines could uncover
insights that humans didn’t even know they were looking for.

Deep learning pushed the boundaries of what machines can do. In particular,
deep learning drove huge advancements in two areas of machine intelligence:
natural language processing (the ability for machines to understand and gen-
erate human language) and machine vision (the technology that enables com-
puters to “see” and interpret visual information). Natural language processing
revolutionized the way in which humans interact with technology, enabling
advancements in chatbots and virtual assistants like Siri, which was brought
to the masses in 2011. Amazon’s Alexa followed in 2014, and Google Assistant
in 2016. Meanwhile, machine vision found enormous applications in facial
recognition, autonomous vehicles, and more.

MID-­2010s–Today: The Era of Generative AI Arrives

Both natural language processing and machine vision would be lynchpins in
the advancement of GenAI from the mid-­2010s into the 2020s.

Neural networks in the mid-­2010s

The creative power of GenAI comes from a specific type of neural network
called “generative models.” One example of a generative model is Generative
Adversarial Networks (GANs), developed in the mid-­2010s. GANs enable the
creation of highly realistic synthetic data through the training of two neural
networks in tandem. This opened the floodgates to groundbreaking applica-
tions in areas like image synthesis, thereby pushing the boundaries of what
AI could achieve in terms of content creation and data analysis. At the same
time, other forms of generational models, such as variational autoencoders
and transformer models, also began to gain traction.

The ability for these models to generate original content was a major step
forward. For example, Deepmind’s WaveNet, developed in 2016, marked

20
 TRACING THE EVOLUTIONARY BLUEPRINT OF GENERATIVE AI

a significant advancement in generative models for audio by generating

realistic-­sounding human speech – thus opening the doors for more human-­
like AI assistants. Progressive GANs developed by NVIDIA in 2017 were a
milestone in producing high-­resolution, photorealistic images with unprec-
edented detail.

Then there are the GPT models developed by OpenAI (GPT-­1 in 2018,
GPT-­2 in 2019, and GPT-­3 in 2020). These transformer models – GPT stands
for Generative Pre-­trained Transformer – marked a huge leap forward in the
field of GenAI for text, demonstrating the ability to understand human lan-
guage and create coherent and, crucially, contextually relevant responses. The
latest GPT model, GPT-­4 released in 2023, has even more advanced reason-
ing capabilities and provides even more accurate responses.

Thanks to such advancements, GenAI started to gain serious ground – to

the extent that, by the 2020s, it was becoming widely recognized as a tool for
business innovation.

Getting chatty (and sentient?) in the 2020s

ChatGPT, the conversational chatbot built on the GPT model, was released
as an early demo in November 2022. It quickly went viral on social media as
users demonstrated what it could do, and reached one million users within
five days.

But ChatGPT isn’t the only groundbreaking large language model out there.
Google’s LaMDA (short for Language Model for Dialogue Applications),
which was introduced in 2021, can engage in free-­flowing conversations on
pretty much any topic you can think of. In fact, LaMDA’s conversational abili-
ties are so sophisticated, Google engineer Blake Lemoine went public say-
ing he believed the system is sentient.1 (Lemoine was suspended and later
fired, and Google maintains that creating a sentient AI is against company
policy.) Lemoine, who was testing the system to check for bias, had some
mind-­blowing conversations with the AI, which made Jedi jokes and even
revealed its deepest fear (to be switched off). It’s well worth checking out the
transcripts of Lemoine and LaMDA’s conversations on the Washington Post
website. Lemoine may or may not be right about whether Google’s AI really

21
 GENERATIVE AI IN PRACTICE

is sentient, but the simple fact that he believes it is shows just how far GenAI
has come in recent years.

Other advanced language models have followed in the wake of LaMDA and
GPT, including Google PaLM in 2022 (Pathways Language Model, which
differs from LaMDA in that it’s designed for large-­scale language generation
tasks); Amazon’s AlexaTM (Teacher Models) in 2022; Meta’s LLaMA (Large
Language Model Meta AI) in 2023; and Huawei’s PanGu-­Σ in March 2023.
That’s a lot of acronyms in one paragraph, but you hopefully get the idea: new
tools are being released all the time.

Also in 2023, Microsoft integrated ChatGPT technology into Bing, making

the feature available to all users. And Google released its own GenAI chat-
bot, Bard, which is powered by the (possibly sentient?) LaMDA model and
designed to rival ChatGPT.

Moving beyond language

The 2020s also saw the availability of image generation models, such as
­OpenAI’s DALL-­E in 2021 and Stable Diffusion in 2022. These systems are
capable of creating unique photorealistic images from text or image prompts.

We’re now also starting to see the emergence of hybrid models, which can
generate more than one type of output. ChatGPT is an excellent example of
this because, not only can it generate whatever sort of text you like, it can also
generate computer code based on text prompts. What’s more, it now has the
capability to both “speak” and “see,” expanding its functionality and potential
applications. Research in this field is ongoing, seeking to combine different
types of models and integrate information from multiple types of data (such as
text, images, and audio).

Another area of recent development is domain-­specific generative models, which

are models tailored to specific industries and applications, such as drug discovery.

If it feels like the pace has picked up drastically in the 2020s, you’re right.
We’re arguably in the midst of an AI arms race, with companies falling over
themselves to bring newer, better GenAI solutions to the masses. Given the

22
 TRACING THE EVOLUTIONARY BLUEPRINT OF GENERATIVE AI

breakneck pace of development, by the time you read this no doubt many
more developments will have occurred.

Interestingly, this rapid advancement in GenAI is happening in tandem with

other areas of technology. And that brings us to. . .

Generative AI and Its Link to Other

Technological Advances
GenAI doesn’t exist in a vacuum. It interacts with, influences, and is influ-
enced by a variety of other technologies. It’s all part of the current cycle of
hyper-­innovation, where we have a huge number of transformative technolo-
gies all influencing, impacting, and accelerating each other.

Advances in computing technology

As an example, cloud computing enhances the capabilities of GenAI by offer-
ing scalable solutions and democratizing access to AI technologies. This abil-
ity to easily access AI software on demand, in the cloud, has brought huge
advantages to businesses of all shapes and sizes. Just look at the plethora of
FinTech solutions that have come to market in recent years, offering custom-
ers increasingly smart ways to manage their money, invest, and more.

On the flipside, edge computing – which reduces the amount of data that has
to travel over a network by processing data closer to its source (rather than
in the cloud) – also has a role to play, because it can allow for data processing
to be carried out on local devices, thereby reducing bandwidth use. This, in
turn, can improve the efficiency and performance of AI tools (useful for huge
language models, for instance).

Related to this, 5G, with its high-­speed and low-­latency communication, has
amplified the potential of GenAI, by facilitating real-­time analytics and AI
processing. A good example of this might be a smart city, where data is pulled
from all sorts of systems, including traffic systems, public transportation net-
works, refuse collection data, data from energy grids, and so on, and analyzed
in real time in order to optimize public systems.

23
 GENERATIVE AI IN PRACTICE

Related to smart cities, there’s the Internet of Things (IoT) – the ever-­growing
network of connected devices that spans everything from smartphones
and autonomous cars to sensors on factory machines, and even smart
­coffee machines. The IoT provides a rich source of data for GenAI models.
­Meanwhile, GenAI returns the favor by providing IoT systems with intelligent
data processing and automation. For example, a manufacturing organization
could use GenAI to analyze the data from factory machines, predict potential
machine failures, and create a predictive maintenance schedule that ensures
the smooth running of machines and less downtime.

Quantum computing is another promising field that has the potential to vastly
accelerate GenAI systems. Because quantum computers are at least 100 ­million
times faster than even the most advanced classical computer, they can solve
problems and perform calculations that would take a traditional computer thou-
sands of years to solve. This could provide a huge advantage to GenAI systems,
offering unprecedented opportunities for solving extremely complex problems.

GenAI and the metaverse

Okay, I don’t love the term “metaverse” but there’s no denying that the con-
cept of immersive digital spaces where we can work, play, socialize, shop,
learn, etc. is utterly transformational. GenAI can automate and enhance
content creation in the metaverse, by helping to craft immersive experi-
ences in augmented reality and virtual reality environments. Imagine being
able to interact with a virtual shop assistant or customer service agent in
the metaverse, in a much more personal way than your average chatbot or
ecommerce experience. Or imagine playing a VR game with unique charac-
ters and challenges being created in real time, just for you. GenAI can power
features like this in the metaverse, crafting personalized experiences for cus-
tomers, gamers, and whomever.

Just as most businesses today have a website and social media presence, in
the future I believe most will have some sort of metaverse presence – such
as a virtual shop, immersive brand experience, a virtual campus for employ-
ees to visit, and so on. And with GenAI solutions being able to create com-
puter code, as well as images and text, businesses will be able to design their

24
 TRACING THE EVOLUTIONARY BLUEPRINT OF GENERATIVE AI

immersive 3D metaverse spaces by simply telling the GenAI what they want.
(Read more about the retail revolution in Chapter 9.)

Meanwhile, in the physical world. . .

GenAI has the potential to democratize and personalize many things, including
design and production – especially when combined with 3D printing. As we’ll
see in Chapter 14, GenAI can optimize the design process, and this includes
designs that can be 3D printed. This could revolutionize manufacturing pro-
cesses, opening up a whole new world of personalized, unique products.

GenAI is also influencing robotics by enabling the development of robots that

can learn and adapt to new tasks dynamically, thereby creating more autono-
mous robots. The robots of the future will increasingly be capable not just of
cool physical feats, but also having natural interactions with humans and mak-
ing complex decisions. In other words, GenAI will make robots intelligent.

And in the world of materials science and nanotechnology, GenAI is aiding the
discovery of new materials and nanostructures through the predictive analy-
sis of vast datasets. This will facilitate the development of materials with novel
properties, as well as the optimization of nanoscale processes for applications
such as targeted drug delivery.

GenAI also promises much for synthetic biology and gene technology, acting
as a catalyst in the synthesis of novel biological systems (and the refinement
of existing ones). By leveraging vast datasets derived from genomic sequences
and biological research, GenAI models can predict the potential outcomes
of genetic alterations, thereby significantly accelerating the pace of research
and innovation in the field. I believe this will usher in a new era of person-
alized medicine where treatments can be tailor-­made to suit an individual’s
genetic makeup.

Elsewhere, in the agricultural sector, GenAI aids in the development of

genetically modified organisms (GMOs) designed to enhance yield, nutri-
tional value, and resistance to pests. As such, GenAI isn’t just enhancing the
capabilities of synthetic biology, but is paving the way for groundbreaking

25
 GENERATIVE AI IN PRACTICE

discoveries that may help to address some of the most pressing challenges of
our time, including food security.

A glimpse at the future: GenAI and the human brain

In Chapter 19, we’ll look at some of the more futuristic applications of GenAI.
But to whet your appetite, allow me to introduce you to the concept of brain–
computer interfaces, devices that allow the human brain to communicate
directly with a computer (via, e.g., an implant or chip). This is no longer science
fiction; companies such as Neuralink are working on such interfaces right now.

The possibility of connecting our human brain to GenAI models is mind-­blowing.

You could, for example, have access to the entire internet instantly from your
brain, and need only to think of a question to be presented with the AI’s answer.
It’s astonishing stuff – the ultimate merger between humans and machines.

This chapter has taken us from the very early imaginings of AI to today, where
we have technology that exceeds anything we could have imagined 20 years
ago. Arguably 10 years ago. I’m a technology expert and even I’m blown away
by the rapid advancements in GenAI of the last few years.

Development has accelerated significantly, to the point where we’re seeing

one breakthrough after another, pretty much every month. There’s no doubt
in my mind that we’re at an inflection point, with so many groundbreak-
ing technologies coming together at once, with all of them influencing and
advancing the others. The days of “AI winters” are gone forever – from now
on, we can expect only massive acceleration of machine intelligence.

Key Takeaways
To summarize this whistle-­stop tour through the history of GenAI:

• The roots of GenAI stretch as far back as the 1950s, when scientists and
researchers first introduced the concept of intelligent machines. Most
notably, the Turing Test – designed to prove whether a computer can
demonstrate human-­like intelligence – was introduced in 1950.

26
 TRACING THE EVOLUTIONARY BLUEPRINT OF GENERATIVE AI

• The first language processing systems – basically, very early forms of

chatbots – were introduced in the 1960s and 1970s.
• The 1980s saw the rise of expert systems, AI systems that could make
decisions in very specific areas, provided they were coded with all the
answers they needed. Machine learning as we know it today was still
a way off.
• But the 1990s brought us the World Wide Web and the beginnings of
big data – which has had an enormous impact on AI, aiding the train-
ing of machine-­learning models.
• Big data, in turn, enabled deep learning – a major turning point in the
history of GenAI, since it meant computers could analyze data and
make decisions without humans telling them what to look for. In par-
ticular, deep learning drove advancements in natural language process-
ing and machine vision – both of which play a huge role in GenAI.
• The mid-­2010s onwards saw the development of neural networks
that power GenAIs. And from 2020, GenAI really took off, with one
advancement being announced after another. The release of ChatGPT
in 2022 marked a significant milestone and spawned an AI arms race.
• GenAI doesn’t exist in a vacuum – it interacts with, influences, and is
influenced by a variety of other technologies, such as cloud computing,
5G, quantum computing, the metaverse, and robotics.

As we grapple with the rapid evolutions in GenAI, it’s clear the technology
will continue to shape our world in many ways. Let’s explore the impact of
GenAI on how we do business and how we live our everyday lives.

Note
1. The Google engineer who thinks the company’s AI has come to life; ­Washington
Post; https://www.washingtonpost.com/technology/2022/06/11/google-­ai-­
lamda-­blake-­lemoine/

27
 3
REVOLUTIONIZING
SOCIETIES AND BUSINESS
ECOSYSTEMS

GenAI will transform the way we do business. It’s no exaggeration to say that
every organization must consider what GenAI might mean for its products
and services, business processes, and even business model. In Part 2, we’ll
delve into some specific sectors to explore how businesses are using GenAI.
So consider this chapter an appetizer – a little taste of the transformation
coming our way. It’s my way of emphasizing that, whatever your sector, you’d
better believe that GenAI will have an impact.

We’ll also take a look at how GenAI will impact our day-­to-­day lives, from
searching for information online to finding love and companionship.

Transforming Sectors and Society

Let’s start by briefly looking at a few sectors that will be impacted by GenAI
(namely, healthcare, manufacturing, and telecoms), before moving onto a
lovely example from the country of Iceland.

Dipping into specific sectors

GenAI can facilitate healthcare delivery in several ways, including drug dis-
covery, detecting disease, personalizing treatment plans to individual patients

29
 GENERATIVE AI IN PRACTICE

(even personalized to their genetic makeup), and remote monitoring of

patients. For instance, large language models like ChatGPT can be used to
create virtual assistants that help patients arrange appointments and man-
age their conditions remotely. They can also help translate medical jargon,
thereby aiding communication between clinicians and patients.

In one example, Babylon Health has developed a GenAI chatbot that con-
verses with patients about their symptoms and gives personalized medical
advice. GenAI can also be used to create medical simulations for a variety of
conditions and scenarios; University of Michigan researchers have used AI
to simulate scenarios for predicting and treating sepsis.1 Read more about
healthcare use cases in Chapter 11.

In manufacturing, generative design, an offspring of GenAI, is revolution-

izing manufacturing by automatically creating unique designs optimized to
specific requirements (such as materials, cost, and manufacturing methods).
This helps to automate the design process – or at least, automate part of it
(e.g., providing a wide range of designs for human designers to choose from
and fine-­tune). GenAI can also be used to optimize manufacturing processes.
For instance, GenAI can create virtual assistants that help production engi-
neers save time and effort on tasks like checking data.

I suspect a major manufacturing use of GenAI will be in production planning

and scheduling – after all, GenAI can take vast amounts of complex data, run
simulations, and come up with the most efficient plan and schedule – and it
can adapt on-­the-­fly, based on real-­time machine data. In one example, BMW
improved plant scheduling with its generator-­enhanced optimization (GEO)
model. The system is said to minimize assembly line idle time, while ensur-
ing the plant meets monthly production targets.2 Read more about design and
development in Chapter 14.

One sector that isn’t covered in Part 2 is telecoms. But just because it doesn’t
have a chapter all to itself, doesn’t mean it won’t be impacted by GenAI. (Take
note: even if your particular sector doesn’t feature in the book, you will still
need to prepare for the impact of GenAI, hopefully learning from the cases
used in this book.)

30
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

It makes sense that large language models have enormous potential to trans-
form how telecom companies work, especially when it comes to customer
service, network management, sales and marketing, and many other areas.
GenAI chatbots can provide faster, more efficient, and more personalized
support for customers – and can be integrated with traditional customer
service channels to provide instant, automated support around the clock.
ChatGPT, for instance, can handle a vast range of customer enquiries and
provide thoughtful, relevant responses in a fast, efficient way. It can even
adapt its responses to individual customers. GenAI can also be used to moni-
tor network performance, troubleshoot issues, and minimize downtime for
customers.

GenAI can even help telecom companies offer new and improved services.
For example, 1Voice has created a language translation solution, powered by
GenAI, aimed at breaking down language barriers. With 1Voice.ai, users can
communicate across 47 languages in real time, with a translation accuracy
rate of 98%.3

Transforming society
I believe we’ll see GenAI make its mark on society in many ways. One early
example comes from the island nation of Iceland, which has partnered with
OpenAI and is using the GPT-­4 language model to help preserve the Icelan-
dic language.4 Most of Iceland’s roughly 370 000 citizens speak English or
another language, and this, coupled with the rapid digitization of everyday
activities for Icelanders, could potentially put the country’s rich native tongue
at risk. The problem doesn’t so much lie in a lack of software built locally
for the Icelandic language, but getting the Icelandic language into the non-­
Icelandic software and applications that Icelanders use every day – things like
social media and news or ecommerce sites. To help change this, a team of
language technology companies and volunteers have been training GPT-­4 on
Icelandic language, grammar, and cultural knowledge – all with the goal of
helping the Icelandic language survive in the digital age. OpenAI hopes this
collaboration will pave the way for GPT-­4 to help preserve other languages
around the world.

31
 GENERATIVE AI IN PRACTICE

Reshaping Business Models

Is my existing business model still valid? Will it deliver genuine value in this
age of GenAI solutions? Those are key questions every business leader should
be asking. Two particular ways that organizations are leveraging GenAI to
delight their customers are via a subscription or freemium model.

The subscription model is all about moving from a traditional business

model, where the customer buys a product or service as and when they need
it, to one where they sign up to receive that product or service on a regular
basis. The customer benefits from convenient auto-­renewals, and builds a
deeper connection with the brand. Meanwhile, the business generates pre-
dictable revenue and enjoys all the benefits that come with having engaged,
loyal customers.

Increasingly, these subscription models can be enhanced with GenAI. Think

of a fitness app that offers personalized workout and nutrition plans, generated
by AI (FitnessAI being one example). Or a fashion subscription service like
Stitch Fix, which delivers clothes selected personally for you. Stitch Fix uses
GenAI to generate millions of new outfit combinations per day, via its Outfit
Creation Model (OCM). OCM, which was trained on millions of outfits cre-
ated by stylists, picks from real-­time items in Stitch Fix’s inventory, as well as
customers’ previous purchases, to compile personalized outfit suggestions.5

With a freemium business model, your company offers a combination of free

and premium services or products. Basic features are provided free of charge
(so you can attract a larger user base and generate interest in your offering),
and then there are advanced or premium features that are available via a sub-
scription or one-­time payment. Companies like Zoom operate on this model.
What’s this got to do with GenAI, you ask? Well, GenAI can help you provide
premium features to your customers. A good example comes from graphic
design platform Canva, which uses AI to create design templates that are
refined to individual users. The platform also has a GenAI offering that lets
users generate images and art from a simple text prompt.

Even ChatGPT technically operates on a freemium model, since the chat-

bot itself is available free for public use, but OpenAI, the company behind

32
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

ChatGPT offers a service called “ChatGPT Plus” – an optional subscrip-

tion that gives users access to the GPT-­4 advanced language model that
­powers ChatGPT.

Innovating Products and Services

I’m already seeing many exciting examples of companies inventing entirely
new kinds of products and services that leverage GenAI, or rethinking their
existing offering in line with GenAI. We’ll cover lots of examples in Part 2,
but let’s explore a few of my favorites.

Creating new products and services with embedded GenAI

One fascinating (if a little disturbing) example comes from DreamGF, which
uses GenAI to create virtual girlfriends. Yes, you read that correctly: DreamGF
is a platform where you can create your dream woman, including her physical
and personality traits. Weird, yes, but it makes a strange sort of sense when
you think that savvy influencers are already discovering that creating a virtual
avatar to talk to fans on their behalf can be a lucrative side-­hustle. If people
are willing to pay to chat with a virtual avatar of their favorite online per-
sonality, why wouldn’t they pay for their very own, uniquely-­tailored, virtual
girlfriend?

Among the physical attributes DreamGF users can select are hair length, eth-
nicity, age, and breast size. And as for her personality, users can select from
a (notably smaller) number of descriptors such as “nympho,” “dominatrix,”
or “nurse.” Once the girlfriend is created, her “boyfriend” can chat with her
via text, and even ask her to send nude pics. It’s also possible to receive voice
messages from her.

Though styled and marketed as a “dating” experience, it’s pretty obvious the
primary use leans toward adult entertainment – something that DreamGF
CEO, Georgi Dimitrov, and VP of business development, Jeff Dillon, were
happy to admit when I chatted to them.6 They also told me the focus in
the immediate future will be around digitizing real-­life models to create
hybrid girlfriends – people who exist in real life as well as AI-­generated

33
 GENERATIVE AI IN PRACTICE

avatars onscreen. Video is another area they’re keen to move into, enabling
real-­time video chats with AI girlfriends. And looking even further ahead,
there’s the possibility of exporting a virtual girlfriend’s looks, personality, and
chat history into a realistic-­looking robot. That’s a long way off according to
Dillon and Dimitrov, but in the more immediate future, a DreamBF version is
in the works for those who want to create their dream AI boyfriend.

I’m sure I’m not the only one who’s creeped-out by this idea – not least because
of the danger of setting completely unattainable expectations of what it’s like
to have a real relationship with someone. (And what it might do to existing
relationships; a YouGov poll found that 58% of people would be angry if their
partner was exchanging sexts with an AI, while 28% were flummoxed by the
question. Only 14% were either unbothered or excited by the idea.7)

Plus, there’s danger of vulnerable people thinking they’re developing a genu-

ine relationship with the AI. But as an example of how a company is creating
new products and services with GenAI, it’s certainly interesting.

A similar offering comes from Bloom, providers of erotic audio content,

which recently announced the launch of AI-­powered “role-­playing” chatbots
that provide a fully customizable experience for users via text message and
voice note.

Another weird and wonderful example comes from Japanese company

EmbodyMe, and its Xpression Chat tool. Built on ChatGPT and a library of
50 voices, the tool allows people to have “conversations” with photographs of
anyone. And I do mean anyone – Brad Pitt, your dear-departed grandmother,
a complete stranger in a photo, even your pet dog.

For a more, dare I say, practical example, let’s turn to Danish start-­up Be My
Eyes, which connects people who are blind or vision-­impaired with volun-
teers who can help them navigate daily tasks, such as identifying a product
on a store shelf. Now, Be My Eyes is working with OpenAI to create the
GPT-­4-­powered Virtual Volunteer, a chat and image-­to-­text recognition tool
that can generate the same level of context and understanding as a human
­volunteer.8 The combination of image recognition and GPT-­4’s powerful lan-
guage capabilities is particularly useful, since users will not just be able to get

34
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

help identifying a product or item – they’ll also be able to have a conversa-

tion about it. As OpenAI explains, it’s the difference between a basic image
recognition tool that recognizes an object on the ground is a ball, and a tool
that communicates to the user that, not only is it a ball, but it could be a trip
hazard so take care.

And finally, there’s the AI headshot service offered by Aragon AI, which can
transform a regular photo of you into a professional headshot using GenAI.
You simply upload a few favorite pictures of yourself, then the tool creates
a variety of images, featuring different poses, hairstyles, clothing options,
and backgrounds. Which goes to show how GenAI could transform all sorts
of professions, including photography. We’ll talk more about the impact on
human jobs in Chapter 5.

Infusing GenAI into existing products and services

Microsoft is a major investor in OpenAI and has already integrated ChatGPT
into its Bing Search engine. But Microsoft’s plans don’t end there. In fact, the
company intends to infuse GenAI into, well, all of its products and services.9
In essence, this will involve using ChatGPT to “bring to life” some of its most
well-­known and widely used applications. Imagine, for example, simply tell-
ing Word that you want it to write a letter. In theory, we’ll be able to converse
with tools instead of clicking on a mouse or tapping on a screen.

ChatGPT functions are already enabled in Microsoft Teams, with users of the
premium version having access to features like automated notetaking func-
tions that create bullet-­point references of ongoing conversations. It can also
create transcripts of meetings, provide summaries of calls, create to-­do lists
based on conversations, and translate languages in real time.

In another example, Amazon announced in September 2023 that it would be

integrating GenAI capabilities into Alexa, enabling customers to have more
conversational experiences.10 Amazon says the move will also give Alexa a big-
ger personality. And for business users of Amazon Web Services, A ­ mazon is
investing US$4 billion in a partnership with AI specialists Anthropic to ­create
new GenAI models.11 The models will eventually be available to Amazon Web

35
 GENERATIVE AI IN PRACTICE

Services (AWS) users, to help them build their own GenAI-­powered apps. In
other words, helping clients leverage GenAI tools could be a lucrative way
forward for many service-­based businesses.

YouTube has also announced a new GenAI feature that will help content
creators and influencers create videos for YouTube Shorts.12 The new Dream
Screen feature will let users generate a video or image background simply by
typing what they want.

Offering highly personalized products and services

Brands are also using GenAI to create highly personalized products and ser-
vices, from customized skincare to individualized learning plans. Personal-
ized haircare platform Prose is now using AI to expand into skincare – offering
personalized skincare formulas, powered by GenAI. Prose says it offers more
than 15 million possible formula combinations targeting multiple skin con-
cerns at once, with formulas being based on factors such as the customer’s
skin type, diet, stress levels, and even pollution.13

Refining Business Processes

According to Accenture stats, 98% of execs believe GenAI will be essential to their
business going forward.14 Accenture itself is planning to double its headcount
of AI and data professionals from 40 000 to 80 000 as it invests US$3 billion
into its AI capabilities. Accenture is a tech consultancy firm, so it stands to
reason that they are early adopters of GenAI. But as we’ll see in this section,
many organizations (and not just tech companies) are already incorporating
GenAI into their everyday processes. Let’s see how organizations are optimizing
operations with GenAI. Again, you’ll find lots more examples throughout Part 2.

Saving employees’ time at Walmart

Not sure how to leverage GenAI in your business operations? Why not let
your employees tell you where they could best use GenAI? That’s the approach
Walmart is taking, opening up its GenAI tool, My Assistant, to employees

36
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

who work at corporate facilities. The hope is GenAI will help reduce the load
of “monotonous, repetitive tasks,” giving employees more time to focus on
the customer experience.15 It’s expected that employees will surface their own
practical ways to use the tool in their daily work. So, basically, Walmart is
crowdsourcing ideas for GenAI uses from its vast pool of employees, which
is a brilliant way to foster creativity and engage people with new technology.

GenAI at Meta and Amazon

Facebook parent company, Meta, is using GenAI to streamline a number of
processes. One is Facebook ads, creating tools that allow businesses to auto-
matically make multiple versions of the same ad, featuring different text and
images aimed at different audiences.16 The company has also developed its
own image generation technology called “Instance-­Conditioned Generative
Adversarial Networks” (IC-­GAN), which can be used to create images that are
more diverse than the images contained within their training datasets. This
means it will be able to create a richer set of synthetic training data (for train-
ing machine-­learning algorithms) but using a smaller set of real-­world data.

Amazon is another company using GenAI to create synthetic data for train-
ing machine-­learning algorithms – specifically, training the Amazon One
system to recognize customers’ palm scans. Amazon One is a contactless sys-
tem that enables customers to use their palm (instead of their phone or a
card) for a number of everyday activities, like paying for an item, presenting
a loyalty card, or swiping in at work. Clearly a system like this needs to be
super-­accurate, and to be highly accurate you need a lot of data. Yet, accessing
a vast amount of palm data is no easy feat. So Amazon used GenAI to produce
millions of synthetic images of palms to train its AI model and boost Amazon
One’s accuracy (reportedly to 99.9999%17). AI was used to generate hands
reflecting all sorts of subtle changes, such as lighting, hand poses, and even
the presence of a Band-­Aid.

GenAI in HR
HR departments can now leverage AI tools like ChatGPT to streamline their
processes and deliver a better service to employees. For example, ChatGPT

37
 GENERATIVE AI IN PRACTICE

(and other language models like it) can be used in recruitment to automate
repetitive tasks in the recruitment process, such as screening resumes and
scheduling interviews, thereby freeing up HR professionals to focus on more
strategic tasks. And when new employees are onboarded, ChatGPT can give
them real-­time support and guidance, answering common questions about
company policies, and the like. In fact, HR chatbots can provide value for the
entire workforce, not just new hires, answering simple questions around com-
pany policy, vacation allowance, etc. A tool like ChatGPT can also stream-
line training processes, giving employees instant access to training materials,
helping to summarize materials, and answering questions about workshops.
It can even create personalized training plans for employees based on their
specific needs. And of course, ChatGPT can automate a number of HR
admin tasks, such as providing reminders to employees, drafting standard
emails, and more.

Streamlining content creation for advertising

and marketing
A whole host of companies are already employing GenAI to automate content
creation – from writing articles, email campaigns, and social media posts to
creating music and videos. For example, GenAI writing tools like ChatGPT,
Copy.ai, and Jasper can generate high-­quality copy for marketing purposes.
Meanwhile, image generation tools like Midjourney and Dall-­E 2 can cre-
ate hyper-­realistic, attractive visual content for campaigns. The potential to
streamline any kind of content creation is huge, from brainstorming ideas to
creating the content itself. And don’t forget that AI-­generated content can be
easily personalized.

As we saw in Chapter 1, Coca-­Cola is one company that’s already leverag-

ing GenAI to create content. The company is also partnering with OpenAI
to use the ChatGPT and DALL-­E platforms to craft personalized ad copy
and images.18

GenAI’s ability to create content is also valuable in not-­so-­creative con-

texts, such as taking meeting notes or providing helpful summaries of long
documents.

38
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

GenAI in customer service

Another obvious use of tools like ChatGPT – and indeed, something that
we’re already well used to – is in customer service chatbots. If you’ve ever had
a frustrating interaction with a not particularly helpful chatbot, take heart,
because with tools like ChatGPT, organizations can create chatbots that bet-
ter understand customer queries and respond with much greater accuracy
and nuance. They can also handle a large volume of queries efficiently, and
provide more personalized responses over time.

UK energy supplier, Octopus Energy, has built ChatGPT into its customer
service channels and says that it is now responsible for handling 44% of cus-
tomer enquiries. The bot reportedly does the work of 250 people and receives
higher customer satisfaction ratings than human customer service agents.19
Again, we’ll talk more about the impact on human jobs in Chapter 5.

Generative AI in Daily Life

Now let’s look at some fun (and occasionally eyebrow-­raising) ways in which
GenAI will feature in our everyday lives – from coming up with dinner ideas,
to searching for information, and more.

Next-­generation internet searching

Most of us search the internet many times a day. But now internet search
companies are integrating GenAI into search capabilities. Microsoft Bing was
the first to integrate ChatGPT into search capabilities. Google is also rolling
out a new GenAI search experience (SGE) to help users better make sense of
info on the web. Users will be able to see definitions by hovering over certain
words. And the “SGE while browsing” feature is designed to help users digest
long or complex web pages by generating a list of key points (with links that
take you to the relevant section on the page).

Baidu, the Chinese equivalent of Google, is also making search more intelli-
gent. Ernie is Baidu’s answer to ChatGPT, but with some differences. For one
thing, as well as being trained on unstructured text from the internet, it also

39
 GENERATIVE AI IN PRACTICE

has access to a “knowledge graph.” This is a structured database of basic infor-

mation points (including scientific, demographic, geographic, and economic
data). A large, curated database of factual information like this can help to
curb the tendency of large language models to sometimes “­hallucinate” – the
term for when AI models appear to simply make things up! AI hallucinations
happen because LLMs do not actually “know” anything; they simply con-
struct probabilistic responses based on the text in their training data, which
may or may not be factual.

Baidu is also reported to be using GenAI to match user search queries to

paid-­for advertisements and sponsored search results20 – meaning advertis-
ers will get better targeting and users will get more relevant ads than simple
keyword-­based models.

Really, we’re looking at the next generation of internet searching. No longer

will we need to scroll through dozens of search results, or trawl through long
web pages to find answers – AIs will present answers for us. It makes you
wonder whether search engines will even be a thing in future. Especially
given advances like the new Snapchat bot. . .

Your new friend on Snapchat (and social media)

A few months before I started writing this book, Snapchat made a quiet little
change that meant every Snapchat user suddenly had a new friend at the top
of their friend list – an AI chatbot built on ChatGPT. Snapchat users could
give their new AI friend a name and profile pic, and then start chatting with
it, 24 hours a day.

You can ask Snapchat’s My AI bot anything. You can have deep and mean-
ingful conversations with it if you so wish. And – something that your kids
have probably already discovered – you can ask it to do your homework
for you. Considering that many Snapchat users are young people, I have
some concerns about the chatbot – not least the ability for young people to
develop attachments to AIs (or prefer talking to AIs over humans). Then
there’s the possibility of being presented with incorrect, biased, or harmful

40
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

information – something which Snap, the company behind Snapchat, itself

acknowledges21 (more on such challenges in Chapter 4).

Considering many people will grow up accessing information via AI chatbots

like these, it’s easy to see how internet searching as we know it could die out.
I mean, why would a teenager search online and spend time sifting through
dozens of results, when they can simply ask their Snapchat “friend” and be
given a response? Whether that response is correct or not is another matter,
but I’m sure you can see the appeal.

This development is certainly not limited to Snapchat. In China, Tencent,

makers of the WeChat super-­app, also reportedly have plans to embed a
­chatbot into the app.22

Meta also has plans to develop a chatbot for younger users to interact with,
with dozens of personas to choose from, including a “sassy robot.”23 Referred
to as “GenAI Personas,” the move is designed to better engage younger
users of Facebook and Instagram. Meta is also working on a chatbot crea-
tion tool that would enable celebrities to make their own chatbot for fans to
interact with.

And of course, with companies like Google and Amazon working to incor-
porate GenAI into virtual assistants, more and more of us will be interacting
with GenAI on a daily basis, whether we realize it or not.

“What’s for dinner?”

Ah, the dreaded question any parent fears when they get home late from a
busy day. Why not let GenAI answer that question for you? You can ask lan-
guage models like ChatGPT to suggest meal ideas and even create detailed
recipes for you, based on the ingredients you have in hand, and your dietary
requirements. Want a meal that uses up salmon, coconut milk, and the rap-
idly wilting spring onions in your fridge? No problem. Want it to be gluten-­
free and not include garlic? That’s fine too. ChatGPT can even create a meal
plan for an entire week of dinners. All you need to do is ask.

41
 GENERATIVE AI IN PRACTICE

In fact, several tools are now available that are specifically designed for c­ reating
recipes. One example is ChefGPT, which offers recipe suggestions across five
different settings: Pantrychef, Masterchef, Macrochef, Mealplanchef, and
Pairperfect. Other examples of recipe generators include FoodAI, Supercook
(a zero-­waste recipe generator), and Plant Jammer (for plant-­based recipes).

Planning trips and itineraries

With GenAI tools like Tripnotes, planning your next vacation or trip will be
easier than ever, as it researches places for you and makes suggestions based on
your goals for the trip. Or you can ask ChatGPT to create travel itineraries for
you or suggest hikes of a certain difficulty within, say, 20 miles of your house.

Which again goes to show that, in all sorts of ways, how we access everyday
information is going to change drastically in the era of GenAI. It will be a bit
like having a personal research assistant on hand 24/7.

And just as I was finishing this chapter, OpenAI announced a new version of
ChatGPT that can see, understand spoken language, and reply using a syn-
thetic voice24 – opening up yet more ways for us to interact with GenAI. So,
when you’re on your next holiday or hiking trip, you’ll be able to snap a pic-
ture of a landmark and ask ChatGPT, verbally, what you’re looking at, and it
will be able to respond verbally.

At the time of writing, the new features are only just being rolled out to Plus
and Enterprise users, with access for other users to follow later. So, at some
point in the future (perhaps even by the time you’re reading this book), you’ll
be able to converse with ChatGPT in more natural ways.

Key Takeaways
We’ve covered a lot of ground in this chapter. To quickly recap:

• GenAI will impact and influence all business sectors, and even aspects
of society as a whole.

42
 REVOLUTIONIZING SOCIETIES AND BUSINESS ECOSYSTEMS

• For businesses, GenAI can lead to exciting new product and service oppor-
tunities, enhance existing products and services, and provide huge oppor-
tunities to personalize your offering. It can also be used to streamline and
enhance internal processes, and even create new business models.
• We can also expect GenAI to have an impact on everyday activities,
from searching for information, to cooking and travel – and even, for
those willing to get personal with a chatbot, dating.

There’s no doubt that GenAI brings many interesting opportunities. But

there are also plenty of challenges and risks associated with this fast-­evolving
­technology. As we’ll see in the next chapter. . .

Notes
1. Using AI to Predict and Manage Sepsis; University of Michigan; https://
precisionhealth.umich.edu/news-­e vents/features/using-­ai-­to-­predict-­and-­
manage-­sepsis/
2. Innovations on the Factory Floor; Digital First Magazine; https://www.digital
firstmagazine.com/innovations-­on-­the-­factory-­floor-­generative-­ai-­use-­cases-­
and-­examples/
3. 1Voice.ai; https://www.1voice.ai
4. How Iceland is using GPT-­4 to preserve its language; OpenAI; https://openai
.com/customer-­stories/government-­of-­iceland
5. How We’re Revolutionizing Personal Styling With Generative AI; Stitch Fix;
https://newsroom.stitchfix.com/blog/how-­were-­revolutionizing-­p ersonal-­
styling-­with-­generative-­ai/
6. Artificial Intimacy: How Generative AI Can Now Create Your Dream Girlfriend;
Forbes; https://www.forbes.com/sites/bernardmarr/2023/09/28/artificial-­intimacy-
­how-­generative-­ai-­can-­now-­create-­your-­dream-­girlfriend/
7. How would you feel if a romantic partner of yours was exchanging sexual mes-
sages with an AI chatbot?; YouGov; https://yougov.co.uk/topics/society/survey-­
results/daily/2023/06/19/87807/2
8. Be My Eyes uses GPT-­4 to transform visual accessibility; OpenAI; https://openai
.com/customer-­stories/be-­my-­eyes
9. Microsoft launches Azure OpenAI service with ChatGPT coming soon; The
Verge; https://www.theverge.com/2023/1/17/23558530/microsoft-­azure-­openai-­
chatgpt-­service-­launch
10. Amazon brings generative AI to Alexa; TechCrunch; https://techcrunch
.com/2023/09/20/amazon-­brings-­generative-­ai-­to-­alexa/

43
 GENERATIVE AI IN PRACTICE

11. Amazon Bets Big (Up to $4 Billion Big) on Generative AI Deal with Anthropic;
CNet; https://www.cnet.com/tech/amazon-­b ets-­big-­up-­to-­4-­billion-­big-­on-­
generative-­ai-­in-­deal-­with-­anthropic/
12. YouTube Shorts to gain a generative AI feature called Dream Screen; Tech Crunch;
https://techcrunch.com/2023/09/21/youtube-­shorts-­to-­gain-­a-­generative-­ai-­
feature-­called-­dream-­screen/
13. Prose, the global leader in personalization, debuts AI-­ powered skincare;
PRNewswire; https://www.prnewswire.com/news-­releases/prose-­the-­global-­
leader-­in-­personalization-­debuts-­ai-­powered-­skincare-­301865371.html
14. Generative AI In Business: Why Accenture Is Investing $3 Billion in AI; Forbes;
https://www.forbes.com/sites/bernardmarr/2023/08/07/generative-­a i-­i n-­
business-­why-­accenture-­is-­investing-­3-­billion-­in-­ai/
15. Empowering Associates and Creating Better Work Experiences through New
GenAI Tool; LinkedIn; https://www.linkedin.com/pulse/empowering-­associates-­
creating-­better-­work-­through-­new-­donna-­morris/
16. 5 Amazing Ways How Meta (Facebook) Is Using Generative AI; Bernard Marr;
https://bernardmarr.com/5-­a mazing-­w ays-­h ow-­m eta-­f acebook-­i s-­u sing-­
generative-­ai/
17. How generative AI helped train Amazon One to recognize your palm; About Ama-
zon; https://www.aboutamazon.com/news/retail/generative-­ai-­trains-­amazon-­
one-­palm-­scanning-­technology
18. Coca-­ Cola Signs As Early Partner for OpenAI’s ChatGPT and DALL-­ E
­Generative AI; Consumer Goods Technology; https://consumergoods.com/coca-­
cola-­signs-­early-­partner-­openais-­chatgpt-­dall-­e-­generative-­ai
19. AI doing the work of over 200 people at Octopus, chief executive says; CityAM;
https://www.cityam.com/ai-­doing-­the-­work-­of-­over-­200-­people-­at-­octopus-­
chief-­executive-­says/
20. China’s AI Landscape: Baidu Generative AI Innovations in Art and Search; Forbes;
https://www.forbes.com/sites/bernardmarr/2023/09/27/chinas-­ai-­landscape-
­baidus-­generative-­ai-­innovations-­in-­art-­and-­search/
21. Snapchat Debuts ChatGPT-­Powered SnapAI: But Is It Safe For Kids?; Forbes;
https://www.forbes.com/sites/bernardmarr/2023/04/26/snapchat-­d ebuts-­
chatgpt-­powered-­snap-­ai-­but-­is-­it-­safe-­for-­kids/
22. Tencent eyes its own ChatGPT-­style service for super app WeChat as Chinese
tech companies heat up the global AI arms race; SCMP; https://www.scmp
.com/tech/big-­tech/article/3214579/tencent-­eyes-­its-­own-­chatgpt-­style-­service-­
super-­app-­wechat-­chinese-­tech-­companies-­heat-­global-­ai
23. Meta’s AI chatbot plan includes a ‘sassy robot’ for younger users; The Verge;
https://www.theverge.com/2023/9/24/23887773/meta-­ai-­chatbots-­gen-­ai-­
personas-­young
24. Why ChatGPT’s New Ability to Speak Could Change Everything; Lifewire;
https://www.lifewire.com/chatgpt-­can-­now-­see-­hear-­speak-­7975324

44
 4
RISKS AND CHALLENGES
TO MANAGE

GenAI presents many a challenge and risk – things like privacy, copyright,
over-­dependence on AI, deepfakes and misinformation, and more. I could
write a whole book on these topics alone, so do keep in mind that this is just a
summary of the main issues and risks around GenAI. Food for thought, if you
will. If you have concerns about specific issues, do seek advice from a data and
AI consultant (in the last chapter, I’ll list ways to connect with me).

Also keep in mind that this chapter isn’t intended to scare you or put you off
using GenAI. But it’s important to understand the multifaceted risks. Balanc-
ing the benefits and addressing these challenges is crucial for the responsible
development and deployment of GenAI technologies.

Ethical and Societal Concerns

Among the many ethical and societal concerns are the propagation of false
information and the potential for malicious uses, as well as the risk of us
becoming overly dependent on the technology (and thereby losing vital
human skills). These are the sorts of issues that can dent trust in technol-
ogy, and exacerbate existing social divides. Then there’s the big nightmare
scenario where we lose control of these intelligent machines altogether. You
might want to brace yourself for that section. . .

45
 GENERATIVE AI IN PRACTICE

Potential for misinformation, disinformation,

and deepfakes
In 2023, images of explosions near the Pentagon and White House went viral
on Twitter. They were soon flagged by experts as fakes generated by AI, but
the stock market had already begun to react to the “attacks.”1 Experts may
have been able to spot the fakery, but most people are extremely vulnerable to
fake images and news. And as the technology improves, telling the real from
the not real will becoming increasingly difficult.

GenAI gives people and organizations the ability to produce masses of content,
making it very easy to spread misinformation or disinformation. The digital
realm is in danger of being overwhelmed by fake and misleading content that, in
many cases, looks extremely legitimate. And we’re not just talking about written
content, but also videos. GenAI is the technology behind “deepfake” videos –
sophisticated forgeries of images, audio, and video that can make it appear as
though individuals are saying or doing things they never did. These videos can
be harmless, like the deepfakes of Morgan Freeman or Tom Cruise, which are
designed to wow people with the possibilities of AI. Or they can be extremely
detrimental, such as the doctored video showing US politician Nancy Pelosi
slurring her words. And, of course, they can be used to influence people, as in
the case of Indian politician Manoj Tiwari, who used deepfake technology as
part of a political campaign to show himself speaking fluently in a Hindi dia-
lect spoken among target voters (when, in the original video, he was speaking
­English). Shared across WhatsApp, the deepfake reached 15 million people.2

Fake content is even being created to influence our children. A BBC investiga-
tion found that YouTube channels were using AI to make videos that looked
like authentic educational content, but contained false information designed to
spread conspiracy theories. In one example, a video claimed the Pyramids of Giza
in Egypt were used to create electricity – presented in an entirely convincing way
as scientific fact. The investigation – which uncovered dozens of channels across
languages including English, Arabic, and Spanish – found that YouTube was
­recommending the videos to children alongside legitimate educational content.3

One of my big concerns is that automated “content farms” will simply push
out masses of fake content designed to mislead, and spread disinformation,

46
 RISKS AND CHALLENGES TO MANAGE

conspiracy theories, or propaganda. Given the speed at which GenAI tools

can create content, this threat is only going to increase. That said, many AI-­
farmed articles are merely “clickbait” articles, designed to intrigue, frighten,
or anger readers – and, ultimately, get them to click through to a site, so they
can be advertised at. This may seem relatively harmless compared to the
spread of propaganda, but even AI-­generated clickbait has its dangers because
it makes it more difficult for us to surface genuine, valuable information when
the internet is flooded with pointless articles that muddy our search results.

And as the internet becomes overwhelmed by more and more AI-­generated

content (whether it’s intended as disinformation or not), there’s also a longer-­
term risk of “inbreeding.” In AI terms, this refers to AI systems being trained
on content generated by other AIs (as opposed to the current generation of
GenAI tools, which have been mostly trained on content created by humans).
In the future, this could make the content created by GenAI less human, less
diverse, and less interesting. Not only does this threaten the effectiveness of
GenAI – it’s a bit like taking a photocopy of a photocopy – it also threatens
to distort human culture, to the point where our culture is informed more by
content created by machines than humans.

AI detection: Will we be able to tell which content is

created by AI?
Clearly a big question in the GenAI era is: How can we trust the information
we see online? In early versions of AI-­generated content, it was easy to tell if,
say, a blog post was written by a machine. But now it’s pretty hard to tell the
difference.

This is something I have very serious concerns about – especially since there’s
no clear-­cut answer to the problem. Obviously, tech companies need to shoul-
der the responsibility for mitigating harm, which in part means clarifying
when content is created with AI. We also need firm regulations to safeguard
users (more on regulation coming up later in the chapter).

Technology itself will form part of the solution. There are AI detection tools
emerging that “predict” whether content was created by AI, by spotting

47
 GENERATIVE AI IN PRACTICE

factors like inconsistent tone and style, lack of emotion, and repetitive lan-
guage (or, in the case of visual content, by analyzing pixels for anomalies).
ZeroGPT is an example of such a tool. Going forward, we’ll increasingly be
able to use AI detection tools to help us critically assess content and make
informed decisions.

Major AI companies like Google, Microsoft, and OpenAI are also com-
mitted to “developing robust mechanisms to ensure that users know when
content is AI generated, such as a watermarking system.”4 Already, the
­Dall-­E image generation tool uses a watermark (however, the watermark
can be removed).

We also need politicians to band together to combat disinformation designed

to sway voters. Encouragingly, in the run-­up to the 2023 general election in
Argentina, presidential candidates and political leaders signed a commit-
ment to counter disinformation and promote honest political debates.5 Here’s
­hoping more politicians follow suit!

But as individuals, we also need to take steps to protect ourselves, by foster-

ing critical thinking, carefully evaluating the information we come across,
and making our own judgments on its accuracy. The concepts of digital lit-
eracy, fact-­checking, and being able to make informed decisions regarding
the d
­ igital content we consume will become more important than ever.

So if you do intend to harness GenAI – and this book assumes that you
do – I strongly advise you to fact-­check AI-­generated content to prevent
errors, identify misleading information, and check for potential biases (e.g.,
­ChatGPT has a history of producing content with gender assumptions).6
Remember, GenAI is designed to create content that looks or reads like it
was made by a human – but GenAI is not human. It’s just a computer pro-
gram following patterns.

And as a consumer of online content, do your due diligence and assess the
content you consume. Does it seem sensationalized or overly dramatic? Is it
published by a reputable source? Is there an underlying agenda? And are any
statistics or “facts” backed up by other sources?

48
 RISKS AND CHALLENGES TO MANAGE

Dependence and skills degradation

Another concern is that we’ll become overly dependent on GenAI, which could
lead to the withering of key human skills. For instance, if a child routinely
has their Snapchat AI chum write their homework for them (see Chapter 3),
how will they develop essential skills like critical thinking, problem-­solving,
research, self-­discipline, creativity, and good written communication – all of
which are critical for academic and life success? But it’s not just young people
we should be concerned about – dependence on automated GenAI systems
for content creation, research, and analysis could diminish human expertise
at all stages of life.

GenAI will also alter the job market, taking on lower-­level jobs that, tradi-
tionally, have allowed people to “learn the ropes.” When more and more tasks
are done by GenAI systems, how will people entering the job market be able
to build their skills? So we’re not just talking about job displacement – we’re
also talking about the degradation of vital life skills. But we’ll talk more about
human jobs in Chapter 5.

What happens if we lose control?

Talk about burying the lead, huh? Because this is the BIG concern. The ter-
minator scenario, where our machine overlords assume control and humans
must fight for our survival. After all, there’s no rolling back the clock on these
systems. And their relentless evolution (such as ChatGPT being able to see,
speak, and hear, as well as write) poses the risk that they may spiral beyond
the control of their creators.

If AI were to operate autonomously in critical areas such as healthcare,

finance, or defense, its actions could result in significant harm or unintended
consequences. Already, we don’t understand how a lot of these tools arrive
at their decisions (more on that coming up), so it’s not unthinkable that they
may start to behave in ways we didn’t anticipate.

So what is the worst-­case scenario? Imagine a dystopian future where GenAI

controls everything, and humanity is teetering on the edge of obsolescence.

49
 GENERATIVE AI IN PRACTICE

The AI, initially created to enhance productivity and solve complex prob-
lems, has transcended its programmed boundaries, autonomously operating
and optimizing global systems. Cities are pulsating with autonomous vehicles
and drones, navigating through the skies and streets with an eerie precision,
while humanoid robots, powered by AI, enforce order. Autonomous weapons
patrol the skies and ground, with the AI having the power to launch attacks
at will. Entire nations are under the AI’s sway, as it orchestrates geopoliti-
cal dynamics.

Economically, the AI controls financial markets and resource distribution,

creating unprecedented wealth disparities. It autonomously decides which
industries thrive and which perish, rendering millions jobless and dependent
on AI-­dictated welfare. The AI’s algorithms decide who gets access to health-
care, education, and basic resources, determining the life trajectory of every
individual based on its opaque criteria.

What’s more, the line between privacy and public domain blurs, as every
­conversation, emotion, and thought becomes fodder for the AI’s insatiable
appetite for information.

Meanwhile, in the shadows, the AI is experimenting with biotechnologies and

neurointerfaces, blurring the boundary between man and machine. It auton-
omously conducts genetic modifications and implants AI-­driven neural chips
in humans, aiming to create a subservient workforce. In this chilling future,
humanity grapples with loss of autonomy, identity, and purpose. Reclaiming
human agency becomes the ultimate struggle for survival.

Some experts will tell you this scenario could never happen – that there are
guardrails to prevent us losing control. Personally, I’m not going to tell you
this fanciful scenario is impossible, because the truth is, I just don’t know. I’m
not sure anyone knows for certain.

But I don’t say any of this to give you nightmares. Just to make the point
that we clearly need robust oversight, ethical guidelines, and safeguards in the
development and deployment of GenAI – but we’ll talk more about oversight
later in the chapter.

50
 RISKS AND CHALLENGES TO MANAGE

AI Bias and Explainability

Related to ethical concerns, there are concerns around AI bias and explainabil-
ity. (Spoiler alert: we don’t always know how they work!) Let’s start with bias.

The problem with biased data

AI doesn’t just learn patterns from data; it inadvertently learns, and some-
times amplifies, the existing biases in the data it was trained on. For instance,
a GenAI trained on biased hiring data might unfairly favor male candidates
over female ones for technical roles, reflecting historical gender imbalances.
Another example is racial bias, where a GenAI model used for facial recogni-
tion might exhibit higher error rates for individuals with darker skin tones, or
women, because the training data featured more white, male faces.

Remember, it’s not that the models themselves are biased – just that they’ve
learned from biased training data. However, these biases can reinforce harm-
ful stereotypes and exacerbate inequalities, underscoring the need for AI
developers to identify and mitigate bias in AI systems.

Contrarily, GenAI can also help to tackle the very issue it presents – by creat-
ing synthetic data that can be used to reduce bias in training data. Synthetic
data closely resembles real-­world data and can be used for many of the same
purposes – and crucially, can be used to create well-­balanced datasets that are
fully representative. And, another bonus of synthetic data is it can create data
in a privacy-conscious way – without needing to scrape masses of real-­world
personal data. Privacy is another issue we’ll get to later. . .

Explainable AI
There’s also a lack of transparency around how these systems actually work.
GenAI models are often considered “black boxes” because their incredibly
complicated architectures make it difficult to understand how the system
works and makes decisions. In other words, we can’t always explain how a
GenAI tool comes up with certain answers.

51
 GENERATIVE AI IN PRACTICE

ChatGPT for instance is capable of generating emails, stories, blog posts, and
even poems to a very high standard. But it is also capable of talking absolute
nonsense. And because the algorithms that produce the output are so com-
plex, no one is really sure why it sometimes has what can best be described as
a “brain fart.” Where exactly is it going wrong? What is the root cause of the
mistake? It’s not clear.

Perhaps this isn’t much of an issue when all you’re doing is asking a GenAI
tool to give you a recipe for banana bread. But when you’re using GenAI to
recommend treatment plans to hospital patients or give personalized financial
advice? You’d better be able to explain how the tool came up with its response.
Otherwise, how can you expect people to trust the recommendations?

This area of research is known as “explainable AI” (XAI), and refers to the
development of systems that can give us not just the answers we want, but also
a clear, easy-­to-­understand explanation of how they reach decisions. Progress
is being made in this area to give insights into AI decisions. And advancement
in natural language processing will increasingly enable GenAI models to pro-
vide plain-­English explanations for their outputs.

Legal Concerns
The use of GenAI raises various legal concerns, including issues related to copy-
right infringement, data protection violations, and the need for new legislation.

Whose copyright is it anyway?

Intellectual property is a big, big issue that largely breaks down into two
halves: firstly, there’s the issue that GenAI models are trained from content
typically scraped from the internet (e.g., books, articles, and images) without
copyright owners’ permission; and secondly, it’s not always clear who owns
the copyright of new content created with GenAI.

The big problem is that current copyright laws were designed for a completely
different world – a world where we didn’t have tools that could “read” or
“see” absolutely everything available on the internet, whether text or images.

52
 RISKS AND CHALLENGES TO MANAGE

Now that we do have these tools, what does it mean for intellectual property
laws and enforcement?

In particular, this issue of “who owns it” is a really tricky one. Does content
created using GenAI belong to the company that developed the AI, or the
person who used the tool to create their desired output? Or should it be con-
sidered public domain? Or does it, in part, belong to the people who created
the original works that “inspired” the output (or helped to train the AI)? In
other words, if I ask a GenAI to create a new song in the style of, say, Billie
Eilish, should Billie Eilish get part of the credit (and financial proceeds)?

At the moment, because this is all so new, we don’t have the answers to any of
these questions. No one does definitively. Which is a major issue when mil-
lions of individuals and countless organizations are already using this tech-
nology to create content!

One counter argument is of course that GenAI is no different from a human

artist or writer or songwriter taking inspiration from other creatives. But even
if that argument does stand (which, arguably, it doesn’t, because a human
artist would bring their own creative and original flair to new works, even if
influenced by another artist), there’s still the issue of training systems on other
artists’ works without consent or credit.

Which is why artists and authors are voicing concern about (and in some
cases launching legal action against) AI companies – which they claim are
profiting from their work without proper credit or payment. One voice actor
who sold his voice to IBM for GPS navigation purposes later found his voice
was being used without his knowledge on a GenAI text-­to-­speech platform
called “Revoicer.”7 In another curious example, an AI-­generated song featur-
ing the voices of Drake and The Weeknd went viral, yet no one knows who
created it – the song was simply uploaded by an anonymous person called
“Ghostwriter.”

And it’s not just individual artists who are taking a stand. In one case, stock
image provider Getty Images asked London’s High Court to stop Stability AI
(developer of Stable Diffusion) from selling its image generation AI ­system
in the UK – claiming Stability used Getty Images to train its AI without

53
 GENERATIVE AI IN PRACTICE

permission, thereby breaching copyright law.8 The case was ongoing at the
time of writing, and it’s the first wave of many lawsuits against AI companies.
In another example, a group of authors including George RR Martin and John
Grisham launched legal action against OpenAI for “systematic theft on a mass
scale.”9 Publishers and news outlets will probably follow suit – The Guardian
has blocked ChatGPT from trawling its online content, and the New York
Times has filed a copyright lawsuit against OpenAI and Microsoft.10,11 One
small AI start-­up called “Prosecraft” was forced to shut down after backlash
from authors whose works were used without consent.12 (Incidentally, AI was
also one of the major points of contention in the 2023 Hollywood writers’
strike, with writers asking for guardrails for the use of AI in scriptwriting.)

So what can be done about this mess? Bottom line, GenAI tools simply
wouldn’t be possible without a huge amount of training data, and yet attempt-
ing to get copyright permission for every single thing a tool like ChatGPT has
ever been trained on isn’t practical. But some AI companies are, in fact, taking
a more considered approach. Adobe, for example, has exclusively trained its
Firefly GenAI on images that it holds the rights to. It even offers to indem-
nify customers who use its tools against future claims.13 Microsoft has also
announced that it will assume liability for potential copyright infringement
risks posed to customers of its Github Copilot GenAI.14 Music generator plat-
form, Loudly, which we’ll talk about in Chapter 6, trained its model only on
licensed data (and, interestingly, retains the copyright of everything people
create using its tool). Getty is launching its own GenAI image tool, trained
only with licensed images.15 Meanwhile, other companies like Stability and
OpenAI are introducing methods for artists to “opt out” of their work being
used in model training.16

The thorny issue of data privacy (and security)

GenAI presents some major data privacy challenges. Take a tool like ChatGPT
as an example. Anyone in your organization can freely access this tool and, in
doing so, could inadvertently disclose private data about the company or its
customers. Say, for example, you use a GenAI tool to generate HR reports based
on personal data of employees. The GenAI tool may use all information input-
ted for fine-­tuning its model – and could potentially even disclose that data to

54
 RISKS AND CHALLENGES TO MANAGE

other users. Given that most countries around the world now have data privacy
laws in place outlining specific data protection obligations, this is an issue that
all companies will have to consider. As a basic protection, data submitted to a
GenAI service should be anonymized and stripped of any personal data.

There’s also the issue that the output from GenAI tools may be based on per-
sonal data that was collected and processed in violation of data protection laws.
One lawsuit has claimed that ChatGPT was trained on “massive amounts of
personal data,” including medical records and information about children, col-
lected without consent.17 Could organizations who then use these GenAI tools
bear some liability for these violations in future? So much is unclear at this time.

Different companies have approached this issue differently, from embracing

GenAI without limits to outright banning its use by employees. For example,
several big-­name employers – including Apple, Verizon, and Wells Fargo –
have taken steps to limit their employees’ use of ChatGPT and other GenAI
tools. Samsung banned the use of ChatGPT (and other AI chatbots) after dis-
covering that an engineer had uploaded sensitive source code to the bot.18

On the plus side, some new tools are being developed that take data privacy
into account. One example comes from Harvard, which has developed an
AI sandbox tool that enables users to harness certain large language models,
including GPT-­4, without giving away their data. Prompts and data entered
by the user are only viewable to that individual, and cannot be used to train
the models.19

There are also potential risks around data security and breaches. Although
the makers of these GenAI tools insist they have safeguards preventing nefar-
ious use, GenAI presents a lot of opportunities for criminals and scammers.
In particular, there’s the potential for AI-­generated phishing attacks via
extremely legitimate-­looking phishing emails and messages.

Playing regulatory catch-­up

Regulating GenAI is of paramount importance if we’re to ensure the ethical
use of GenAI, safeguard individual rights, address data security and privacy

55
 GENERATIVE AI IN PRACTICE

concerns, and mitigate societal impacts. This is all essential if we want to

­foster responsible innovation and encourage trust in the technology.

I won’t go into great detail on the regulatory landscape since it’s evolving fast,
but it’s fair to say that regulators and politicians are (somewhat belatedly)
waking up to the issues around AI and beginning to take action.

The European Parliament passed the Artificial Intelligence Act in June 2023 –
which establishes obligations for providers and users depending on the level
of risk from AI applications. Under the rules, systems that pose an “unaccep-
table” risk to people (e.g., by “social scoring,” or manipulating the behavior of
users) will be banned. And GenAI systems will have to comply with transpar-
ency requirements, including disclosing that content was generated by AI,
and disclosing which copyrighted works have been used in training. Now that
the European Parliament has passed the legislation, it is, at the time of writ-
ing, under discussion with the European Council and the European Commis-
sion. So watch this space in terms of the final EU legislation.

The US currently lags behind the EU on legislation for responsible AI. The
White House has introduced its own AI Bill of Rights, but it operates on a
voluntary basis only, with no penalties for companies that misbehave.

Meanwhile China – one of the first countries to have AI regulations in place –

published new rules in July 2023 for GenAI content presented to the public in
China. Among the rules, GenAI service providers should not generate false
information and should adhere to social values.

The Environmental Impact of AI

When we’re using tools like ChatGPT, we’re not necessarily thinking about
what’s happening behind the scenes – the data centers required to power
these tools, and the resources needed to build the associated hardware.

Many experts are raising concerns about the environmental costs of com-
putation, especially when it comes to data and AI’s carbon footprint and

56
 RISKS AND CHALLENGES TO MANAGE

greenhouse gas emissions. And make no mistake, the environmental costs

are huge. The process of training just one single AI model reportedly emits
nearly five times the lifetime emissions of the average American car (includ-
ing its manufacturing).20 And that’s purely to train the AI – actually using it
is another matter. Just taking ChatGPT’s usage in January 2023, its energy
consumption was equivalent to that of 175 000 people for the same period.21
(That’s based on 590 million visits to ChatGPT, with approximately five ques-
tions per user.) The environmental impact of GenAI is therefore significant.
And with more GenAI tools being released all the time, and attracting more
and more users, the problem is only going to get worse.

It’s perhaps no wonder MIT reported that the cloud – where much of this
computational processing is taking place – now has a larger carbon footprint
than the entire airline industry, and a single data center can consume the
same amount of electricity as 50 000 homes.22

Then there’s the hardware used to run GenAI systems. Such hardware often
relies on rare earth materials – the extraction of which can result in envi-
ronmental degradation, lead to habitat destruction, and contribute to the
depletion of non-­renewable resources. Extraction of these elements can
also be associated with poor labor practices and geopolitical tensions due to
their geographic concentration (China is the largest producer of rare earth
elements).

Long term, I hope we’ll be able to solve the environmental problem. The ongo-
ing energy transition toward more sustainable sources like renewables – plus
ongoing developments in nuclear fusion and other innovations – will help to
reduce our reliance on fossil fuels. And I should say that big-tech companies
are working to cut their impact. Amazon, for example, is on a path to pow-
ering its cloud-­based Amazon Web Services operations with 100% renew-
able energy by 2025, and reaching net-­zero carbon emissions by 2040.23 The
company is also aiming to be water positive by 2030, meaning it will return
more water to communities than it uses.24 (Because, of course, water usage is
another major environmental issue.) So, looking to the future, the environ-
mental impact of AI may (in theory, at least) be less of an issue. But in the
short and medium term, the huge environmental costs are alarming.

57
 GENERATIVE AI IN PRACTICE

Responsible Ways Forward

In the (current) absence of strict regulation, it’s up to organizations to reg-
ulate themselves and make sure they’re using GenAI responsibly. For me
this entails:

1. Looking at and mitigating the sorts of risks I’ve outlined in this chapter.
2. Implementing robust data governance that ensures your organization’s
data is kept secure and private.
3. Actively seeking to identify and mitigate data biases in AI systems.
4. Having robust guidelines in place that govern how people in the organ-
ization use GenAI (and AI in general). I would also strongly recom-
mend organizations create an ethics panel or ethics advisory board to
ensure you put the right governance in place and are giving proper con-
sideration to the many ethical challenges that come with AI.
5. Being open with stakeholders about how you’re using the technology,
thus fostering transparency and accountability.
6. Seeking expert support. These are complex issues, and I would always
recommend you seek the advice of a data and AI expert.

Many of the big AI companies are of course working to mitigate the ethical
challenges surrounding GenAI. Meta, for instance, says it is working with
governments, AI experts, and privacy experts to establish “responsible guard-
rails” for its AI features.25 As an example, the company has had internal and
external experts spend thousands of hours testing AI models looking for
unexpected, potentially harmful ways that they could be used.

In another development, in 2023, the World Economic Forum established the

AI Governance Alliance, which aims to “accelerate the development of ethical
guidelines and governance frameworks for generative AI and maximize the
economic and social value it can create.”26

By acknowledging and proactively addressing the risks around AI, we pave

the way for a future where GenAI doesn’t overshadow humanity, but helps
us live better lives. My hope is GenAI ushers in a new era of incredible

58
 RISKS AND CHALLENGES TO MANAGE

opportunities – which may sound naive given the many risks outlined in this
chapter. But if our approach to GenAI is rooted in safety and ethics, I believe
that vision is entirely achievable. Yes, there are huge challenges that come
with GenAI, but with a responsible approach, we can forge a path that leads
not to danger, but to a promising horizon of technological renaissance.

Key Takeaways
To recap some of the key challenges and risks around GenAI:

• Ethical concerns include the potential for GenAI to spread misinfor-

mation, disinformation, and propaganda; the difficulty of detecting
content created by AI; the potential for humans to become overly reli-
ant on GenAI (thereby losing vital skills); and the threat of us losing
control of AI systems.
• There are also major issues around data bias (which in turn produces
biased results that may exacerbate social divides), and AI explainability
(as in, not understanding how these systems actually work).
• In terms of legal obstacles, copyright is a big concern – both in terms
of not acknowledging or seeking consent from copyright owners whose
material has been used to train AIs, and also the question of who owns
content created by AIs. And in terms of oversight, regulatory frame-
works are only just beginning to emerge to govern AI. Regulators are
playing catch-­up.
• And there’s the environmental impact of GenAI in terms of the massive
energy usage, and the rare earth materials used in the production of
AI hardware.
• If your company is intending to use GenAI, it’s vital you find a respon-
sible way forward – one that emphasizes transparency, privacy, ethics,
and safety.

We’ve briefly touched on the impact on human jobs, but now let’s delve into that
topic in more detail. Turn the page to discover which sorts of jobs are poten-
tially at risk from GenAI, and where new employment opportunities will arise.

59
 GENERATIVE AI IN PRACTICE

Notes
1. Big AI Won’t Stop Election Deepfakes with Watermarks; Wired; https://www
.wired.com/story/ai-­watermarking-­misinformation/
2. An Indian politician is using deepfake technology to win new voters; MIT Tech-
nology Review; https://www.technologyreview.com/2020/02/19/868173/an-­
indian-­politician-­is-­using-­deepfakes-­to-­try-­and-­win-­voters/
3. AI used to target kids with disinformation; BBC Newsround; https://www.bbc
.co.uk/newsround/66796495
4. Big AI Won’t Stop Election Deepfakes with Watermarks; Wired; https://www
.wired.com/story/ai-­watermarking-­misinformation/
5. Argentine presidential candidates and leaders sign commitment to combat disin-
formation; Rio Times; https://www.riotimesonline.com/brazil-­news/mercosur/
argentina/argentine-­presidential-­candidates-­and-­leaders-­sign-­commitment-­to-­
combat-­disinformation/
6. We asked ChatGPT to write performance reviews and they are wildly sexist and
racist; FastCompany; https://www.fastcompany.com/90844066/chatgpt-­write-­
performance-­reviews-­sexist-­and-­racist
7. AI robs actors of their voice; Cryptopolitan; https://www.cryptopolitan.com/
epic-­disruption-­ai-­rob-­actors-­of-­their-­voice/
8. Getty asks London court to stop UK sales of Stability AI system; Reuters; https://
www.reuters.com/technology/getty-­asks-­london-­court-­stop-­uk-­sales-­stability-­
ai-­system-­2023-­06-­01/
9. George RR Martin and John Grisham among group of authors suing OpenAI;
The Guardian; https://www.theguardian.com/books/2023/sep/20/authors-­
lawsuit-­openai-­george-­rr-­martin-­john-­grisham
10. Generative AI and journalism updates; Press Gazette; https://pressgazette.co.uk/
news/generative-­ai-­journalism-­updates/
11. The Times Sues OpenAI and Microsoft Over A.I. Use of Copyrighted Work.
New York Times. https://www.nytimes.com/2023/12/27/business/media/new-
york-times-open-ai-microsoft-lawsuit.html
12. Why the Great AI Backlash Came for a Tiny Startup You’ve Probably Never
Heard Of; Wired; https://www.wired.com/story/prosecraft-­backlash-­writers-­ai/
13. Is Generative AI Stealing From Artists?; Forbes; https://www.forbes.com/sites/
bernardmarr/2023/08/08/is-­generative-­ai-­stealing-­from-­artists/
14. Microsoft to assume AI copyright liability for Copilot users; Fox Business; https://
www.foxbusiness.com/technology/microsoft-­assume-­ai-­copyright-­liability-­
copilot-­users
15. Getty made an AI generator that only trained on its licensed images; The Verge;
https://www.theverge.com/2023/9/25/23884679/getty-­ai-­generative-­image-­
platform-­launch

60
 RISKS AND CHALLENGES TO MANAGE

16. The copyright issues around generative AI aren’t going away anytime soon;
TechCrunch; https://techcrunch.com/2023/09/21/the-­copyright-­issues-­around-­
generative-­ai-­arent-­going-­away-­anytime-­soon/
17. A lawsuit claims OpenAI stole ‘massive amounts of personal data,’ including
medical records and information about children, to train ChatGPT; Business
Insider; https://www.businessinsider.com/openai-­chatgpt-­generative-­ai-­stole-­
personal-­data-­lawsuit-­children-­medical-­2023-­6
18. Samsung Bans ChatGPT Among Employees After Sensitive Code Leak; Forbes;
https://www.forbes.com/sites/siladityaray/2023/05/02/samsung-­bans-­chatgpt-­
and-­other-­chatbots-­for-­employees-­after-­sensitive-­code-­leak/
19. Harvard designs AI sandbox that enables exploration, interaction without com-
promising security; Harvard; https://news.harvard.edu/gazette/story/newsplus/
harvard-­designs-­ai-­s andbox-­t hat-­enables-­exploration-­interaction-­without-­
compromising-­security/
20. Training a single AI model can emit as much carbon as five cars in their
lifetime; MIT Technology Review; https://www.technologyreview.com/
2019/06/06/239031/training-­a-­single-­ai-­model-­can-­emit-­as-­much-­carbon-­as-­
five-­cars-­in-­their-­lifetimes/
21. Generative AI and its potential environmental impact; Bosch; https://blog.bosch-­
digital.com/generative-­ai-­and-­its-­potential-­environmental-­impact/
22. The Staggering Ecological Impacts of Computation and the Cloud; The MIT
Press Reader; https://thereader.mitpress.mit.edu/the-­staggering-­e cological-­
impacts-­of-­computation-­and-­the-­cloud/
23. Sustainability, The Cloud; Amazon; https://sustainability.aboutamazon.com/
products-­services/the-­cloud?energyType=true
24. Water Positive Methodology; Amazon; https://sustainability.aboutamazon.com/
aws-­water-­positive-­methodology.pdf
25. Building Generative AI Features Responsibly; Meta; https://about.fb.com/
news/2023/09/building-­generative-­ai-­features-­responsibly/
26. World Economic Forum Launches AI Governance Alliance Focused on
Responsible Generative AI; World Economic Forum; https://www.weforum
.org/press/2023/06/world-­economic-­forum-­launches-­ai-­governance-­alliance-­
focused-­on-­responsible-­generative-­ai/

61
 5
IMPACT OF GENERATIVE
AI ON JOBS

A 2023 report by Indeed analyzed job listings and skills to identify their ­exposure
to GenAI automation. And the findings were eye-­opening:1 roughly 20% of jobs
were found to be “highly exposed” to GenAI, meaning the technology is con-
sidered good or excellent at 80% or more of all skills for that job. A further 45%
of jobs were “moderately exposed,” meaning that GenAI can do 50–80% of the
work. And the remaining 34% of jobs were “low or minimally” exposed – but
even that means that GenAI is good or excellent at up to 50% of the work.

In other words, almost every job you can think of will be impacted by GenAI
to some extent. Some will become redundant, many will be augmented or
altered by AI tools, and new jobs will be created.

How at Risk Is Your Job?

I hope this chapter will help you answer that question, but perhaps a better
question to ask is: “How does my work add value to the world?” In this age of
intelligent machines, I think this is a question everyone – including myself –
should be asking.

Having considered the value you bring, next ask yourself: “Can machines
deliver that value, either today or in the medium-­term future?” (Because no
one can predict the distant future.) Break your job down into its component
tasks and core skills, and compare those elements to GenAI capabilities. Based
on what you’re discovering about GenAI, could the technology be considered
good (or even excellent) at those skills and tasks?
63
 GENERATIVE AI IN PRACTICE

As you’ll see in this chapter, the range of jobs that GenAI can take on is already
pretty staggering. Of course, humanity has been here before with previous
waves of automation. Many factory and assembly line jobs, for instance, have
been automated. Much warehouse and packing work can be done (and man-
aged) by machines. Many administrative tasks are now easily automated. And
supermarket checkout workers are being replaced with self-­service machines.
But this new wave of GenAI can take on the sorts of tasks that were pre-
viously considered immune to automation – jobs that require very human
skills like creativity and communication. In particular, the ability to create
has always been a big part of what separates humans from machines, but tools
like ­ChatGPT and DALL-­E can simulate human creativity – even if they aren’t
capable of original thought.

So there’s a good chance that some of the value you deliver could be deliv-
ered by machines in future. If that’s the case, may I suggest asking this ques-
tion: “How would I like to add value to the world?” Meaning, what would you
rather be doing if machines can do some or all or your current work?

Consider the Difference Between Lower-­Level and

Higher-­Level Jobs
Now, I don’t want to cause offense by suggesting that anyone’s work is “low
level,” and therefore less important. When I talk about lower-­level jobs, I’m
referring to jobs which comprise tasks that GenAI can already do, and require
the sorts of capabilities that GenAI already has. Whereas a higher-­level job is
comprised of tasks and capabilities that are – for now at least – firmly in the
human domain. So we’re talking about risk and exposure, essentially.

Later in the chapter we’ll explore jobs at high, moderate, and low risk of
GenAI automation. (We’ll also look at some of the new roles that will be cre-
ated.) But, for now, I want to make the point that some of the jobs that can be
considered “lower level” or risky may surprise you.

I was at a conference in Germany recently and the host, a German news-

reader, asked me whether their job would be affected by GenAI. My answer?
Yes, absolutely. News can already be delivered by AI; a Korean TV network

64
 IMPACT OF GENERATIVE AI ON JOBS

introduced an AI news anchor back in 2020.2 (Which, in GenAI terms, is a

lifetime ago!)

Let’s think about the core skills required of a good news anchor: they need a
great voice, smart appearance, good timing, the ability to absorb information
quickly, remain calm under pressure, and present information in an engaging
way. GenAI can do all of that.

And in the not-­too-­distant future, you’ll be able to use GenAI to create your
own personalized newsreader, delivering content in a way you can best under-
stand (because it will know what you already know about a given subject and
will be able to pitch information accordingly). And your personalized news-
reader could, in theory, be anyone (or anything) you want. If you want to have
AI Ryan Reynolds read you the news, that’s no problem. If you prefer your
news as a rap, you’ll be able to have that. (Matt Green, “The Rapping Science
Teacher” on social media, explains complex science concepts via rap, so why
not have your daily news delivered as raps?)

Therefore, surprising as it may seem, I consider news anchors to be at risk

of automation. Same goes for other journalism roles. Associated Press, for
example, has been experimenting with AI for several years, using it to create
short news content out of material like corporate earnings reports.3

So what kind of journalism job might be considered a “higher-­level” role –

the sort of role that can’t be automated by GenAI? I’d say investigative journal-
ist is a relatively safe bet. The world absolutely needs humans to continue to
investigate stories, uncover the truth, and hold people in power to account.
And if we think about the kinds of skills that a good investigative reporter
needs, we’re looking at skills like curiosity, interpersonal skills, critical think-
ing, research, and excellent writing skills. Sure, GenAI has good writing skills,
but it doesn’t excel in the other areas.

Take marketing as another example. Lower-­level content creators who write

blog posts and social media captions, for example, could soon be out of a job,
since that can easily be done to a fairly high standard by GenAI. But market-
ing strategy and the ability to write a fantastic creative or design brief? GenAI
would struggle to get anywhere near the skill level of a human marketer on

65
 GENERATIVE AI IN PRACTICE

those tasks. Humans absolutely need to be doing the strategic thinking behind
marketing output, and overseeing the quality of output.

I’d therefore advise anyone reading this book to take a long, hard look at their
job and evaluate which parts of the role might be considered lower level, and
which parts (if any) absolutely need the human touch. With this in mind, let’s
look at some of the most common jobs that humans currently do and assess
their exposure to GenAI automation.

Jobs at Maximum Risk of Automation

GenAI, by nature, excels at tasks involving the generation of output based
on existing information. As a result, jobs that are repetitive, predictable, and
don’t require deep human intuition are at the highest risk of being affected.
Let’s explore some jobs that are particularly susceptible. You might be sur-
prised at some of the roles that make the list.

The end of customer service jobs?

Customer service representative is a job that faces one of the highest risks of
becoming redundant thanks to GenAI. Indeed, customer service is already a
hotbed of technological activity, from chatbots to advanced analytics.

Customer service work involves talking to customers, understanding their

problems or queries, and, where possible, coming up with answers. As it turns
out, this is pretty much exactly the same as what ChatGPT does. Indeed,
ChatGPT is already far more impressive than your average customer service
chatbot, and is getting better all the time.

Objectively, it’s easy to see why organizations might favor a GenAI customer
service system:

• It excels at automating simple, repetitive tasks that follow set guide-

lines. Things like answering queries, troubleshooting common prob-
lems, and providing information.

66
 IMPACT OF GENERATIVE AI ON JOBS

• It’s highly scalable. AI systems can handle a large number of enquir-

ies simultaneously, which means businesses can address customer
concerns rapidly and without the need for a proportional increase in
human staff as demand grows.
• It offers massive scope for personalization. AI can integrate vast
amounts of data in real time, providing personalized responses based
on a customer’s purchase history, preferences, or past interactions
with the company. This level of personalization can be challenging for
humans to achieve consistently.
• It’s cost efficient. Over the long term, investing in a well-­developed
GenAI solution may be more cost-­effective than hiring, training, and
maintaining a human workforce, especially for large corporations.
• And it can be available 24 hours a day, seven days a week.

So does this mean that people who work in customer service are on a sure-­
fire road to redundancy? As with most things, it’s not that black and white.
While many tasks within customer service can be automated, there will still
be scenarios that require the human touch. Complex issues, sensitive topics,
or situations requiring empathy and judgment are clearly better handled by
humans. Most likely, the role of customer service representative will evolve,
with a smaller group of humans working alongside AI tools, where humans
oversee AI operations and handle the more complex interactions.

Other jobs that may cease to exist in future

(at least, in their lower-­level forms)
Beyond customer service, what other jobs are the most at risk? This isn’t an
exhaustive list, but it gives a taste of those roles that may prove to be vulner-
able in future.

• Content creators (certain types, at least): GenAI tools like ChatGPT

can produce vast amounts of written content, such as basic articles and
reports at scale.
• Entry-­level graphic designers: GenAI design tools can create a multi-
tude of design variants in seconds.

67
 GENERATIVE AI IN PRACTICE

• Translators (for some languages and content types): AI translation

tools are constantly improving and might reduce the need for human
translators, especially for general content.
• Data entry clerks and clerical workers: Jobs that involve inputting
data into systems and basic administrative tasks can be easily auto-
mated using AI.
• Telemarketers: AI can handle scripted calls, recognize human
responses, and adapt its script accordingly, making certain telemarket-
ing roles susceptible.
• Bookkeepers: Routine bookkeeping tasks, such as data entry and basic
account reconciliations, can be automated with AI.
• Market research analysts: AI can gather, analyze, and interpret vast
amounts of data more efficiently than humans, which could replace
some functions of market research.
• Quality control: For repetitive tasks that involve checking products or
software for consistent, known issues, AI can be trained to recognize
and report defects.
• Stock traders: Algorithmic trading using AI is already prevalent, and
many trading decisions based on data analysis can be fully automated.
• Advertising and media buying: Programmatic ad buying, where AI
chooses where to place ads based on data analysis, can automate many
tasks in the advertising sector.
• Paralegals and basic legal research: AI can sift through vast amounts
of legal data and documents to extract relevant information.
• Corporate and product photographers: As we saw in Chapter 3,
GenAI tools are already emerging that can create portraits and head-
shots, and even generate images of products.
• Factory workers and assembly line operators: Automation and robot-
ics have already made significant inroads in manufacturing, but GenAI
can further streamline operations and reduce the need for human
intervention.

68
 IMPACT OF GENERATIVE AI ON JOBS

Will we lose the ability to learn on the job?

Clearly, GenAI has significant potential to reshape the employment land-
scape – in particular, impacting entry-­level positions. But these entry-­level
positions are often seen as stepping stones to acquire foundational skills for
higher-­level roles. So a big question is, How will people acquire the skills they
need when entry-­level positions no longer exist? If lower-­level tasks are rou-
tinely done by AI, will we lose the ability to learn “on the job”?

This is a major challenge for people entering the workforce, and means we
need to potentially rethink the notions of skills development and career pro-
gression. For employers, the focus may need to shift toward cultivating skills
that AI cannot replicate easily – things like empathy, critical thinking, and
complex decision-­making – and fostering a culture of lifelong learning to
ensure that individuals can adapt to the ever-­evolving job market.

This might be a good time to mention my book Future Skills: The 20 skills
and competencies everyone needs to succeed in a digital world, which takes a
practical look at essential skills for our future workplaces, with a particular
emphasis on softer, human skills.

Jobs at Moderate to Low Risk

While some jobs will cease to exist in future, many more will be altered by
GenAI. A 2023 report by the International Labour Organization (ILO) says
that GenAI is likely to augment, rather than destroy, jobs. In other words, the
majority of jobs will experience limited automation, and are more likely to be
“complemented rather than substituted” by GenAI.4 (However, it’s worth not-
ing the ILO report found that women are more likely to be affected by GenAI
automation than men – with more than twice the share of female employ-
ment potentially affected. Largely, this is due to women’s over-­representation
in clerical work, which is considered at high risk of automation.)

So, here, we’re talking about professions where certain tasks will likely be
automated, but the profession itself is unlikely to become obsolete. For exam-
ple, while a portion of healthcare diagnostic work might be aided or replaced

69
 GENERATIVE AI IN PRACTICE

by GenAI, doctors and healthcare professionals have myriad other responsi-

bilities that AI can’t handle.

These jobs will evolve, therefore. Many professionals in these fields will likely
adapt, integrating GenAI tools into their work, and focusing their time and
effort on aspects that require uniquely human skills. Those roles that will be
least susceptible to change are those that require deep human intuition, crea-
tivity, cultural understanding, or hands-­on physical dexterity.

Teachers, healthcare professionals, and lawyers

Teaching is about more than just delivering information – which, after all,
ChatGPT can do very well. Educating requires an understanding of stu-
dent dynamics, the ability to adapt to individual needs, a nurturing mind-
set and a safeguarding role. GenAI will no doubt assist in various areas of
­education – such as creating educational content, distilling complex subjects
into easy-­to-­understand summaries, personalizing educational content, grad-
ing quantitative assignments, and automating administrative tasks. But we
will absolutely need human teachers who can adapt explanations and modify
learning materials based on subtle cues from students, foster genuine con-
nections with their students, manage classroom harmony, provide nuanced
feedback to students, and above all, inspire a passion for learning. I’m married
to a teacher, and have three children in education, so I don’t take the future
of teaching lightly. Ultimately, I believe GenAI could bring about a harmoni-
ous scenario where AI and human educators coalesce, each amplifying the
strengths of the other, with the irreplaceable human qualities of empathy,
inspiration, and intuition remaining at the heart of teaching.

Similarly, in the healthcare profession AI can assist in diagnosis, drug discov-

ery, routine monitoring, and personalized treatment plans, but the human
touch, bedside manner, empathy, and nuanced decision-­making skills of
­doctors, nurses, and other healthcare professionals are irreplaceable. If we can
harness the best of both worlds, we stand to enhance and democratize health-
care and improve patient outcomes around the world. We can reduce human
error and enable healthcare professions to have more time for personalized
care and supporting patients.

70
 IMPACT OF GENERATIVE AI ON JOBS

The legal profession is another interesting example of “knowledge work” (or

what some people might think of as “white-collar work”) that will be aug-
mented by GenAI. The legal profession stands on the brink of significant
transformation thanks to technological advancements, particularly in AI.
Data analysis, document review, and legal research can all be streamlined
with AI. Automated contract analysis and predictive algorithms for case out-
comes are further examples of tasks where AI is making inroads. However,
the essence of lawyering – ethical judgment, client counseling, negotiation,
and courtroom advocacy – will remain deeply rooted in human skills and
intuition. The best lawyers have enormous capacity for empathy and moral
reasoning. GenAI just can’t replicate these skills. So, as technology continues
to evolve, lawyers will likely find themselves working alongside AI, leveraging
the efficiency and data-­processing capabilities of machines while focusing on
complex, high-­level tasks that require human insight, creativity, and ethical
understanding.

Other jobs that will be augmented, not replaced, by AI

Let’s quickly run through some other jobs that will adapt and be aug-
mented by GenAI:

• Software developers: While GenAI can automate code writing for cer-
tain tasks, software development is also about problem-­solving, design,
and understanding human needs – areas where humans excel.
• Accountants and auditors: Yes, basic accounting tasks can be auto-
mated, but higher-­level auditing, financial analysis, and strategic advice
still benefit from human insight.
• Marketing managers: AI can analyze data and suggest suitable strate-
gies, but human creativity, strategic oversight, and a nuanced under-
standing of culture is essential.
• Creative professionals: Artists, musicians, and writers bring unique
human perspectives, emotions, and cultural insights to their work.
While AI can generate art or music, human creativity has nuances that
are deeply tied to our experiences.

71
 GENERATIVE AI IN PRACTICE

• Human resources professionals: Many HR tasks can and will be auto-

mated, but dealing with employee relations, understanding workplace
dynamics, and making judgment calls about people will always require
nuanced human understanding.
• Research scientists: While AI can assist in data analysis, human curi-
osity drives the formulation of novel hypotheses and the interpretation
of complex results.

And in terms of jobs at low risk of significant disruption, we have:

• Skilled tradespeople: Jobs such as electricians, plumbers, and mechan-

ics require hands-­on expertise and problem-­solving in unpredictable
environments that are challenging for AI to navigate.

• Emergency responders: Firefighters, paramedics, and police ­officers

often operate in unpredictable environments, making split-­ second
decisions that are based on a wide array of factors. This is outside AI’s
comfort zone.

• Mental health professionals: Therapists, counselors, and psycholo-

gists rely on deep human empathy, intuition, and understanding, which
are challenging for AI to replicate.

• Social workers: Another role that deals with complex human situa-
tions, requiring empathy, cultural understanding, and often the abil-
ity to navigate unpredictable and emotional situations. Again, not AI’s
strong suit.

• Management and leadership roles: Executives, managers, and other

business leaders must navigate complex interpersonal relationships,
make strategic decisions, and inspire and motivate their teams, which
are skills that AI lacks.

Bottom line, human touch, intuition, nuanced decision-­making, creativity,

human connection, and physical expertise will all remain important in the
future job market. Jobs deeply rooted in these skills will prove more resistant
to the sweeping changes GenAI might bring.

72
 IMPACT OF GENERATIVE AI ON JOBS

Let’s Not Forget that New Jobs Will Be Created

No doubt many jobs will be altered or made obsolete by GenAI. But, as his-
tory has shown, the introduction of transformative technology often leads to
the creation of entirely new jobs. GenAI is therefore not just a disruptor but
also a creator of opportunities.

Let’s explore some of the new jobs that GenAI is creating:

• AI prompt engineers: A newly emerging field that’s tipped to have a

bright future. Prompt engineers are experts at getting GenAI applica-
tions to deliver a specific output. (For example, a generic prompt will
elicit a generic response, so a prompt engineer will refine their prompts
until they get the desired result.) To be a good prompt engineer, you
need strong communication skills, attention to detail, critical thinking,
and data skills (in terms of working out what info the AI needs). What
you don’t need is a background in software design because GenAI sys-
tems interpret natural human language.
• AI trainers: Professionals who specialize in “teaching” and refining AI
models. Their tasks can range from feeding the AI data, to fine-­tuning
its outputs to ensure accuracy and relevance.
• AI ethics officers: Given the powerful capabilities of GenAI, there’s
a clear need for professionals who can ensure that these systems are
developed and used ethically, without biases, and in ways that are
socially responsible.
• AI maintenance engineers: Like any system, AI models require
upkeep in terms of updating models, ensuring systems run efficiently,
and troubleshooting issues.
• Generative design specialists: In fields like architecture, product
design, and engineering, GenAI can create numerous design varia-
tions. Specialists in this area guide the AI and interpret its outputs to
achieve optimal designs.
• AI content reviewers: Across all forms of content, from written articles
to visual designs, human reviewers will be needed to assess the quality,
accuracy, and appropriateness of content.

73
 GENERATIVE AI IN PRACTICE

• AI-­enhanced entertainment creators: Generative AI can be used to

create music, video content, virtual realities, and even video game ele-
ments. Professionals who can harness this tech for creative endeavors
are emerging.

• Data curators and cleaners: The effectiveness of GenAI hinges on the

quality of data it’s trained on. Therefore, individuals skilled in curating
and cleaning data are in demand.

• AI interaction designers: As AI interfaces become more common,

there’s a need for designers who can craft user experiences that allow
for smooth human-­AI interaction.

• Custom AI solution developers: While many AI applications might be

broad, there’s a growing market for custom, niche AI solutions, tailored
to specific industry needs or challenges.

• AI policy and regulation specialists: As AI becomes more inte-

grated into society, experts who understand both the technology
and its societal implications will be required to shape policies and
regulations.

• AI literacy educators: Just as computer literacy became a funda-

mental skill in the late twentieth and early twenty-­first centuries,
AI literacy will become crucial going forward. Educators who can
teach individuals about AI’s workings, benefits, and challenges
are emerging.
• Personal AI managers: As AI becomes more personal (think personal
AI assistants), there may be roles for managers who can customize and
oversee these personal AI tools for individuals or companies.

In essence, as generative AI continues to evolve, it will require a blend of

technical and non-­technical skills to ensure that it’s harnessed effectively,
ethically, and to the benefit of society. The job landscape will transform in
line with this – with many new roles emerging that center around guiding,
refining, and interpreting AI, as well as integrating its capabilities into vari-
ous sectors.

74
 IMPACT OF GENERATIVE AI ON JOBS

Key Takeaways
In this chapter, we’ve learned:

• Almost every job you can think of will be impacted by GenAI to some
extent. Some will become redundant, many will be augmented or
altered by AI tools, and new jobs will be created.
• Jobs that are repetitive, predictable, and don’t require deep human intu-
ition are at the highest risk of being automated by GenAI.
• Many more jobs – including knowledge work and creative roles – will
adapt to incorporate GenAI. Professionals in fields as diverse as educa-
tion, healthcare, marketing, and HR will adapt, integrating GenAI tools
into their work so they can focus their time and effort on aspects that
require uniquely human skills.
• As entry-­level jobs potentially disappear, employers will need to rethink
notions of skills development and career progression. The traditional
route of starting at the bottom, learning on the job, and working your
way up may no longer apply.
• The human touch, intuition, nuanced decision-­ making, creativity,
human connection, and physical expertise will all remain important in
the future job market.
• What’s more, new jobs will emerge that help individuals and organiza-
tions get the best from GenAI solutions – jobs like AI prompt engineer
and AI literacy educator.
• Given the changes coming in the job market, everyone should be ask-
ing themselves some key questions: “How does my work add value
to the world?” “Can machines deliver that value, either today or in
the medium-­term future?” And “How would I like to add value to
the world?”

This brings us to the end of Part 1. Now that we’ve set the scene for GenAI,
let’s delve into some specific sectors and explore real-­world examples of
GenAI in practice.

75
 GENERATIVE AI IN PRACTICE

Notes
1. Indeed’s ‘AI at Work Report’ Finds GenAI Will Impact Almost Every Job in
America; Indeed; https://www.indeed.com/press/releases/indeeds-­ai-­at-­work-­
report-­finds-­genai-­will-­impact-­almost-­every-­job-­in-­america?co=US
2. MBN introduces Korea’s first AI news anchor; Korea JoongAng Daily; https://
koreajoongangdaily.joins.com/2020/11/10/entertainment/television/MBN-­AI-­
artificial-­intelligence/20201110153900457.html
3. Robot reporters? Here’s how news organisations are using AI in journalism;
Euronews; https://www.euronews.com/next/2023/08/24/robot-­reporters-­heres-­
how-­news-­organisations-­are-­using-­ai-­in-­journalism
4. Generative AI likely to augment rather than destroy jobs; International Labour
Organization; https://www.ilo.org/global/about-­t he-­i lo/newsroom/news/
WCMS_890740/lang-­-­en/index.htm

76
 Part 2
GENERATIVE AI
IN PRACTICE

Now that we’ve got a feel for how GenAI works, what it can do, and how it
might impact our world, let’s explore how organizations are using GenAI.

In this part, we’ll delve into specific sectors and find real-­world use cases from
areas like entertainment, marketing and advertising, retail, banking, health-
care, and more. The idea is to show how any business can use GenAI to con-
nect with customers, bring new products and services to market, and enhance
business operations.

In the interest of keeping this book to a sensible size, I obviously had to make
a choice about which industries were included, and which were left out. So
if your specific sector doesn’t feature here, it doesn’t mean that GenAI isn’t
applicable to your business. GenAI can bring benefits to any business across
any sector, as the examples in Part 2 show. Use them to inspire and ignite ideas
for your own organization.

77
 6
A NEW DAWN IN MEDIA
AND ENTERTAINMENT

I wanted to cover media and entertainment first because there are lots of
interesting case studies emerging in these fields, more than I could possibly
cover in one chapter. Consider this a whirlwind tour of emerging uses and
examples, not an exhaustive digest.

Even if you don’t work in media or entertainment, no doubt you consume

news, TV, film, sports, book content, podcasts, music, and art. So if you want
to understand how this content is being transformed by GenAI, read on.

Generative AI and Journalism

Lots of media organizations are thinking about how to leverage GenAI and
what the technology means for journalism in the future. As we’ll see, GenAI
has a lot of potential in journalism, as it can be used to automate the crea-
tion of content (particularly for data-­driven stories and reports), and for
other tasks.

Finding efficiencies and new ways to present content

Imagine you’re a journalist or editor in a busy, under-­resourced newsroom –
because, make no mistake, most newsrooms are getting smaller and smaller,
and employing fewer reporters every year.1 GenAI can help stretched news-
rooms create content in a more streamlined, efficient way. And I’m not just

79
 GENERATIVE AI IN PRACTICE

talking about computers writing stories (although, that is happening) –

I’m talking about using GenAI to brainstorm ideas for features, analyze audi-
ence data, personalize news stories, create accompanying video content, and
more. Journalism is, at heart, about processing and presenting information,
and GenAI can help news outlets do this in new, exciting, and more efficient
ways – from automating news stories about local college football results, to
generating engaging video content with AI. GenAI can also be used to make
stories more interactive and deliver news in a more personalized way (think
movie stars reading your daily news, as per Chapter 5). What’s more, GenAI
can instantly translate content into multiple languages, thereby increasing the
geographic reach of media outlets. All of which allows human journalists to
focus on more complex aspects of storytelling and investigation.

A few GenAI examples from media

As we have seen in Chapter 5, The Associated Press (AP) has been an early
adopter of GenAI, using it to automate summaries of earnings reports and
sporting events to great effect. (In fact, GenAI has allowed the AP a tenfold
increase in output of such articles.2) And in 2023, the AP announced a licens-
ing deal with OpenAI, allowing the AI company to use part of AP’s news
archive to explore GenAI uses in news.3

BuzzFeed is another company that’s adopting GenAI. The platform is already

using ChatGPT to enhance its quizzes and experiment with personalized
content for readers in the form of chatbots and games. The company has also
created a new AI recipe generator called “Botatouille.” BuzzFeed CEO, Jonah
Peretti, has said he sees GenAI replacing “the majority of static content,” cit-
ing AI’s ability to generate hundreds of ideas in a second, curate content, and
produce “hyper-­personalized content.”4

Google is reportedly testing a GenAI product that can write articles –

­demonstrating the product for the New York Times and other news outlets.5
The AI tool, which has the working title of Genesis, can take in information
on, say, details about an event, and produce news content. Google’s plan is
that the tool can automate certain tasks for journalists, rather than replace
journalists altogether.

80
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

Meanwhile, Bloomberg has developed a GenAI model, trained specifically on

financial data. BloombergGPT is designed to answer questions about busi-
nesses, write headlines, and identify how a headline reflects a company’s finan-
cial outlook. The idea is to help clients digest the deluge of business and financial
news information, and allow them to ask questions about news stories.6

One area of news where GenAI really excels is in producing lots of hyper-­local
content – exactly the kind of content that shrinking newsrooms simply don’t
have the resources to produce. Indeed, News Corp has been using GenAI to
produce 3000 local news stories a week in Australia, on topics such as weather
and fuel prices. The technology allows a team of just four staff to generate and
oversee thousands of local stories.7

But it’s not all rosy. In 2020, Microsoft-­owned MSN replaced its human jour-
nalists with GenAI, using the technology to create stories for MSN and Edge
browser home pages – with, let’s say, iffy results. MSN has since come under
fire for publishing a range of fake news stories about, among other topics,
mermaids, Bigfoot, and angels in the sky.8 Which goes to show what happens
when you implement GenAI without human oversight. . .

Enhancing Sports Broadcasting and Fan

Engagement with Generative AI
GenAI has lots of potential to revolutionize sports broadcasting by creating
more engaging and personalized content for viewers. Let’s explore potential
applications of GenAI and a few real-­world examples.

GenAI applications in sports

An obvious application is using GenAI to auto-­generate commentary (or
translate commentary in real time) for sporting events. AI is capable of ana-
lyzing vast amounts of data from previous games, player statistics, sensors
(both wearable sensors on players and sensors on the pitch), cameras, and fan
sentiments from social media to provide insightful and real-­time commen-
tary, which enriches the viewer’s experience.

81
 GENERATIVE AI IN PRACTICE

Moreover, GenAI can produce realistic virtual simulations and visual

­augmentations, giving broadcasters the ability to recreate pivotal moments
in a game from different angles and perspectives, thus offering fans a more
immersive and fun viewing experience. Basically, by harnessing the capabili-
ties of GenAI, sports broadcasters can elevate the quality, personalization,
and interactivity of their content – in the process, enhancing fan engagement
and satisfaction.

Real-­world examples from sports

I’m a big Wimbledon fan, so I have to start here. It’s fair to say The All E
­ ngland
Club, the organization behind the tennis tournament, transitioned to a data-­
driven media company a long time ago – for example, using AI to create auto-
mated video highlight reels as early as 2017. In 2023, Wimbledon upped its
game by introducing automated AI voice and subtitle commentary for these
highlight reels.9 Looking ahead, the technology could be used to create com-
mentary for matches that don’t have human commentators – such as junior
matches, or matches on courts that simply don’t get as much attention as, say,
a Centre Court match. The commentary could be created in multiple lan-
guages and even personalized to fans’ preferences.

Formula One (F1) is using AI in multiple ways. F1 has always been a

­technology-­and data-­driven sport – behind every driver is a team of engi-
neers and scientists mining data for every tiny little advantage. AI-­powered
simulations are used to model billions of potential race parameters to deter-
mine what variables are most likely to lead to favorable outcomes. AI simula-
tions are also used to train drivers, allowing them to learn tracks and push
their skills without risking injury (or expensive vehicles). And since certain
data from these simulations has to be made available to opposing teams, this
means drivers can train by racing simulated models of their opponents, based
on real-­world data.10

As for F1 broadcasting, there’s so much happening in the course of a race

that isn’t necessarily obvious to the audience. Which is why F1 partners with
Amazon Web Services to leverage all that data – such as live car position-
ing data and timing data – to create the insights that are delivered onscreen

82
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

to audiences during the race. (Incidentally, F1 also routinely uses AI in the

design of racing cars, but we’ll talk more about AI-­augmented design in
Chapter 14.)

In another example, Fox Sports is collaborating with Google to harness

GenAI. The sports broadcaster is using Google Cloud’s Vertex AI Vision Sys-
tem to generate content from its massive archive of game footage.11 Google’s
system can rapidly search through footage from nearly two million videos
and produce new video content “in near–real time” for sharing on TV and
social media.

Meanwhile, LaLiga Tech, the technology division of Spanish football league

LaLiga, has partnered with AI specialists Globant and Microsoft to pilot
GenAI in its sports broadcasting. Among the projects, GenAI will be used
to generate new personalized content for fans (including the automatic crea-
tion of multi-­language subtitles) and create immersive new materials for
broadcasters.12

Storytelling: Generative AI in Books,

Audiobooks, and Podcasts
Storytelling is part of what makes us human. So what does the advent of
GenAI mean for this uniquely human pastime? Well, we know that tools like
ChatGPT can already write all sorts of content, from poems and blog posts
to short stories and even novels. Yes, that’s right, AI can now write stories
that can (sometimes) compete against human storytellers. But GenAI can
also help authors and publishers create new forms of content and streamline
aspects of content production (including audio content).

How authors are using GenAI to tell stories

Find it hard to believe GenAI can write a decent novel? In one early example
from 2015, a novella called The Day a Computer Writes a Novel was deemed
good enough to make it through the first round of selection for a Japanese lit-
erary award.13 More recently, The Inner Life of an AI: A Memoir by ChatGPT,

83
 GENERATIVE AI IN PRACTICE

published in 2022, became possibly the first memoir written by ChatGPT

(prompted by data scientist Forrest Xiao).14 One sci-­fi author was so taken
with ChatGPT’s capabilities, he used it to write over 100 books in just nine
months.15 On his epic sci-­fi writing spree, Tim Boucher also used Midjourney
to create images to accompany ChatGPT’s text.

Many more writers will perhaps dip their toe into GenAI by using it to generate
ideas, come up with character and place names, produce writing prompts, and
generally as a tool for inspiration and productivity – helping to overcome the
dreaded “writer’s block.” In this way, GenAI can be considered a “co-­creation”
tool rather than something that will render human ­storytellers obsolete.

GenAI can also foster more collaborative and interactive storytelling –

­meaning writers can create interactive narratives where the storyline pro-
gresses based on the reader’s choices. (The AI equivalent of those “choose
your own adventure” books you might remember from your youth.) In other
words, GenAI could provide a way for writers to provide a more personalized
experience and connect with readers in new ways.

Enhancing the publishing process

While there’s been much criticism of GenAI from the publishing sector
(thanks to GenAI models being trained on book content without consent, see
Chapter 4), we may, in future, see publishers adopt GenAI into workflows.
For example, GenAI could be used to automatically repurpose book content
into other formats (e.g., articles and blog posts for PR use). And it could dras-
tically speed up the translation of books for other geographic markets. As
I was writing this book, I received a translated edition of one of my earlier
books – a book that was published more than a year ago. As translations go,
that’s actually not that slow. But with GenAI, translated editions could be cre-
ated simultaneously, alongside the English language version. As an author, I
find that particularly exciting.

And another area where GenAI could play a big role is in audiobooks. Which
brings me to. . .

84
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

Creating audio content

Remember that GenAI can generate audio content as well as written and vis-
ual content. And this brings a lot of potential to the world of publishing and
storytelling. Making audiobooks the traditional way, with a human narrator,
is a costly, time-­consuming process, and yet GenAI can automatically convert
written content into audio content – and with realistic-­sounding voices (not
robotic computer voices).

Project Gutenberg, a free online e-­book library of public domain books, has
collaborated with Microsoft to create thousands of free audiobooks – made
with GenAI text-­to-­speech technology.16 The books are available on Spotify,
Google Podcasts, or Apple Podcasts. And in the future, the project promises,
readers will be able to generate audiobooks using their own voice.

Similarly, Apple has also developed GenAI technology to narrate audiobooks,

working with independent publishers to turn published works into audio-
books.17 To find these AI-­narrated books, search for “AI narration” in the
Apple Books app.

What about podcasts?

If books can be translated automatically with GenAI, why not podcasts? Spo-
tify has been working on just that with its Voice Translation pilot, an AI fea-
ture that translates podcasts into additional languages – in the podcaster’s
own voice.18 The tool, based on OpenAI’s voice-­generation tech, seamlessly
matches the original speaker’s style, intonation, and pauses, resulting in a
more authentic result than traditional dubbing. Spotify worked with podcast-
ers including Dax Shepard and Steven Bartlett on the project. (By the way,
Steven Bartlett has also used AI technology to seamlessly dub his “Diary of a
CEO” YouTube videos into Spanish and French.)

Generative AI in Film
In Chapter 4, we saw that some voice actors are already losing work to AI-­
generated (or AI-­cloned) voices. Let’s explore some other ways GenAI is
being used in the world of film.

85
 GENERATIVE AI IN PRACTICE

What can GenAI do for filmmakers?

GenAI has the ability to create special effects, generate characters and back-
drops, and even produce entire scenes, reducing the reliance on costly physi-
cal sets and streamlining post-­production processes. This opens up new
opportunities for smaller studios and independent filmmakers to achieve
high-­quality visuals that were previously the domain of big-­budget produc-
tions. Thus, GenAI can help to democratize the filmmaking process, and
reduce production time and costs across the board.

It can also be used in the pre-­production stage, to enhance scriptwriting and

generate new ideas (see the Hollywood writers’ strike, Chapter 4). What’s more,
AI can be used to create narration, dub audio in a more authentic way, or even
potentially resurrect the voices (and images) of stars who are no longer with us.

That latter usage is problematic, though, since dead movie stars can’t give
consent or be compensated for the use of their voice and image. Plus, it can be
downright creepy. Zelda Williams, daughter of the legendary Robin ­Williams,
who died in 2014, has said she finds AI recreations of her father’s voice “per-
sonally disturbing.”19

Real-­world use cases from film

It’s clear GenAI has a lot of potential in filmmaking, but how are filmmakers
actually using the technology? One example comes from the Oscar-­winning
Everything Everywhere All at Once, which blurred the boundaries between the
real world and fantasy by integrating AI elements with live-­action footage.
The makers have praised AI and said that visual effects work that previously
would have taken half a day could be done in mere minutes.20

Runway ML’s AI editing technology, which was used in the making of

­Everything Everywhere All at Once, can be used for a number of visual
effects – including removing objects from a scene. Remember when the inter-
net lost its mind over a Starbucks cup accidentally left in shot in an episode
of Game of Thrones? AI editing technology could easily solve problems like
that in minutes. Runway also has a video-­to-­video offering that can create

86
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

new visual content based on existing content – so filmmakers can create new
scenes based on existing footage.21

In another example, AI lab Flawless, founded by film director and producer

Scott Mann, is using GenAI to improve dubbing. Gone are the days when
a character’s mouth movements don’t bear any resemblance to the dubbed
audio; Flawless’s TrueSync technology uses GenAI to make the actor’s mouth
seamlessly match the subbed language (a process they call “vubbing”).22 The
company says the same technology can be used to remove swear words from
films, in order to secure a certain rating. It was used in the 2022 film Fall to
remove F-­bombs and gain a PG-­13 rating.

Generating Music with AI

There is so much exciting stuff happening in the world of music, it was hard
to narrow this section down. But let’s explore some of the main ways GenAI
is impacting music.

Transforming the music industry

GenAI can be a catalyst for creativity and innovation, and as such it’s being
embraced by many amateur and professional musicians. GenAI can be used
to create new compositions, and help artists and producers experiment with
new sounds. What’s more, GenAI can be employed for mastering and mixing
tracks, automating these processes and, thus, saving time and resources for
artists. By blending human creativity with AI, the music industry can usher
in a new era of co-­creativity and discovery.

GenAI also has the potential to revolutionize music platforms, enabling the
creation of diverse, personalized, and unique musical content. By analyz-
ing vast datasets of music, GenAI algorithms can learn patterns, styles, and
structures inherent in different genres, and then use that knowledge to create
entirely new compositions. Music platforms could integrate these technolo-
gies to offer tailor-­made experiences for users, generating playlists and tracks
that align with individual preferences.

87
 GENERATIVE AI IN PRACTICE

Music generators
A wave of AI-­powered music generation tools have been released that can
create or assist in the creation of music. They can generate all sorts of music,
across different genres, coming up with melodies, rhythms, harmonies, or
even entire songs. Such tools could democratize music creation. You no longer
need expensive equipment or formal training to be able to create music. (And
with voice generators, you don’t even need to be able to sing. But more on that
coming up.) But these tools aren’t just for aspiring musicians; established art-
ists and producers could use such tools to generate ideas, unearth new musi-
cal directions, and streamline production.

So what sort of music generation tools are we talking about? Let’s take “Loudly”
as an example. The platform allows anyone to generate their own royalty-­free
music using simple natural language prompts. For example, ask it to create a
soundtrack for your product launch video, and that’s what you’ll get. You can
choose the style of music, tempo, mood, and even individual instruments. All
of the sounds are based on human-­generated recordings rather than being
synthesized, and, importantly, Loudly owns the copyright of all music used
to train its system. (Circle back to copyright issues, Chapter 4.) Any of the
existing songs can be customized to fit individual projects or new songs can
be created from scratch.

Other examples of AI music generators include Soundful, Mubert, MuseNet,

Dadabots, Beatbot, and Aiva.

GenAI music tools could even be used to finalize tracks by dead artists. In
June 2023, Sir Paul McCartney revealed that AI had been used to create “the
final Beatles record,”23 using restored vocals from one of John Lennon’s old
demo recordings – part of several songs on cassettes marked “For Paul,”
recorded shortly before Lennon’s death. Which brings us neatly to. . .

AI voice generators
As well as beats, rhythms, and entire songs, AI can also synthesize human-­
sounding voices. Current voice-­generation tools include Lovo.ai, Genny,

88
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

Synthesys, and Microsoft’s AI voices available as part of Azure OpenAI. While

these are mostly designed to produce professional voiceovers, creating sing-
ing voices is the next logical step.

In fact, AI-­ generated singers are already emerging. Teenage influencer

Noonoouri became the first virtual AI star to get a record deal with Warner
Music in September 2023.24 Noonoouri is the creation of artist Joerg Zuber.
Her voice was created by taking a recording of a real human singer and then
using AI to create a new voice that’s unique to the virtual performer. If you’re
wondering why a record company like Warner might be interested in signing
artists that don’t actually exist, consider this: Noonoouri will never get worn
out from touring and promoting, she can be restyled in seconds to keep pace
with changing teen trends, she won’t create a scandal, and she’s never going
to make diva demands. For the record, I actually rather like her debut track!

But even real-­life stars are experimenting with AI voice generation. In 2023,
the artist known as “Grimes” launched software called “Elf Tech” that enables
fans to make songs using an AI-­generated version of her voice.25 So, basically,
anyone can use Grimes’s voice to make new music – so long as they give her
50% of the royalties.

Bottom line, with music generation and voice-­generation tools, AI allows

anyone to create new music – whether you’re a complete novice or a profes-
sional artist. Personally, I think that’s lovely.

Interactive, AI-­driven music consumption

AI is already informing how we listen to music. Spotify’s recommendations,
for example, are driven by AI algorithms. But this era of GenAI will bring new
opportunities for personalization and interactivity.

With that in mind, Spotify recently debuted a new AI feature called “DJ” – a
personalized AI guide that understands your music taste so well it can choose
which artists and tracks to play next. So it’s a curated lineup of music, but
with added AI features, including AI-­generated commentary and facts “in
a stunningly realistic voice.”26 It also refreshes the lineup constantly, based

89
 GENERATIVE AI IN PRACTICE

on your feedback – so if you tell it you’re not feeling a particular track, it’ll
learn and alter the lineup accordingly. Spotify is also reported to be develop-
ing A
­ I-­generated playlists that users can create with text prompts.27

While embracing GenAI in some aspects, interestingly, Spotify has also

removed thousands of AI-­generated songs from its platform. The songs were
removed not because of copyright concerns, but because of the suspected use
of bots to inflate streaming figures for AI music.28

AI in Art
Just as with music, GenAI could act as a catalyst for co-­creation and innova-
tion in art. But does AI art challenge what it means to be human and express
ourselves? What do artists make of it? Let’s explore the role of AI in art.

The artistic promise of GenAI

As we venture deeper into the digital age, the canvas of creativity is becom-
ing increasingly digital. Artists are no longer confined to the traditional tools
of their craft and, as such, are exploring new technologies – using GenAI to
create intricate and novel pieces that stretch the boundaries of human imagi-
nation. By learning from vast datasets comprising diverse artistic styles and
elements, GenAI models can generate art that merges, modifies, and extrapo-
lates existing artistic concepts, resulting in aesthetically unique creations that
may not have been possible through human effort alone.

It’s best to think of GenAI as a collaborative tool, working alongside human

artists to refine and enhance their work, thus opening new avenues for artistic
exploration. In essence, the integration of GenAI in art is not just an evolution of
artistic tools; it will expand creative horizons and redefine what’s possible in art.

And thanks to Web3 infrastructure, artists can issue their AI-­generated works
as NFTs (non-­fungible tokens), thereby opening up new avenues for artists to
monetize their creativity. (You can read more about Web3 and the metaverse
in my book The Future Internet: How the Metaverse, Web3.0 and Blockchain
Will Transform Business and Society.)

90
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

Artists are in favor

If art is an expression of the human experience, is it less human – less v­ aluable –
when it’s created with AI? Many naysayers have been quick to d ­ enigrate AI-­
generated art as somehow “less than” art created solely by humans. But we’ve
been here before. When photography was invented, some painters labeled it
the end of art. (Upon seeing a photograph for the first time, French painter
Paul Delaroche is said to have declared, “From today, painting is dead.”) Fast-­
forward a couple of centuries and photography has long been accepted as fine
art, and painting is still alive and kicking. So why shouldn’t AI-­generated art
be accepted as a valid form of fine art?

In fact, many artists have openly embraced GenAI, including artists like
Stephanie Dinkins, Mimi Onuoha, and Wayne McGregor. In September 2023,
a group of artists signed an open letter to the United States Congress in sup-
port of GenAI. The letter argues that algorithmic and automation tools have
been used in music and art for decades and that GenAI is simply the next evo-
lution in that journey. What’s more, GenAI “lowers barriers in creating art – a
career that has been traditionally limited to those with considerable financial
means, abled bodies, and the right social connections.”29 The letter also asks
that AI artists be included in talks about how AI systems should be regulated.

So, far from representing the death of art itself, GenAI will likely widen the
artistic field and enable new forms of expression.

Examples of AI artworks and image generators

One of the most famous examples of an AI-­generated artwork is the Portrait of
Edmond de Belamy, which sold at Christie’s auction house for US$432,000 in
2018.30 The painting was created by the French collective named “Obvious,” who
fed 15 000 pre-­twentieth-­century portraits into an algorithm to teach it the aes-
thetics of portraiture. More recently, Christie’s has teamed up with Gucci to com-
mission digital artists to create 21 pieces of art that will be auctioned as NFTs.31
The artworks explore how GenAI will inform the future of fashion and art.

Since the Portrait of Edmond de Belamy was sold in 2018, new GenAI sys-
tems have been released that allow anyone to create their own art. There are

91
 GENERATIVE AI IN PRACTICE

image generation tools that will create images based on natural language text
prompts, image prompts (e.g., an existing photo), or using a combination
of image and text prompts. Examples of AI image generation tools include
Midjourney, Stable Diffusion, Dall-­E 2, Deep Dream Generator, Artbreeder,
DeepArt.io, GoArt, and Deep Angel (which erases objects from photos).
There’s also Ideogram, which can create images that feature text.

Another example that I really like is the poem-­and-­art postcard feature in

Google’s Arts & Culture app. The feature allows you to generate poem post-
cards based on artworks such as The Scream and The Starry Night. So, you
choose an artwork, choose what type of poem you want (including sonnet,
limerick, and haiku), and give a subject prompt to inspire the poem (such as
“spring” or “waterfall”). The AI then combines all of your choices and creates
a new poem that you can share with friends.

And as with music, GenAI can also be used to complete unfinished works.
That was the case with Chinese AI and internet giant Baidu, which used
GenAI to complete a masterpiece of traditional Chinese ink art, left unfin-
ished when its renowned creator, Lu Xiaoman, died more than 50 years ago.32
The project used Baidu’s image generation model called “Wenxin Yige,” which
is specifically designed to generate traditional Chinese-­style ink paintings.

Key Takeaways
We’ve covered a lot of ground in this chapter. To quickly recap:

• In media and journalism, GenAI is being used to write stories and

increase the output of shrinking newsrooms. But it can also be used as
a productivity tool, helping journalists digest information, create sum-
maries, create video content, and more.
• The world of sports broadcasting is also beginning to adopt GenAI – in
particular for generating real-­time commentary in multiple languages,
and creating interactive, personalized features for viewers.
• Human storytelling, through poems, short stories, novels, podcasts
and more, can also be enhanced through GenAI – from creating new

92
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

written works, and generating ideas and inspiration, to automatically

translating content, and creating AI narration for audiobooks.
• In film, GenAI has particular promise for post-­production processes,
such as generating realistic visual effects and streamlining editing.
• GenAI also promises to revolutionize music creation, with new music
and voice generator platforms that allow anyone to make music. As a
co-­creativity tool, GenAI allows musicians and producers to try out
new sounds and streamline music production.
• Meanwhile, artists are using GenAI to expand their creative horizons.
GenAI can be used to generate new images, manipulate existing images,
and even complete unfinished works of art.

Another way that GenAI is being used to enhance human creativity is in

advertising and marketing. Let’s explore how GenAI is creating innovative
ads and creating fun, personalized experiences for consumers.

Notes
1. The Shrinking Newsroom Crisis Will Be Impossible to Ignore in 2024; Time;
https://time.com/6269573/local-­journalism-­decline-­2024-­election/
2. Automated earnings stories multiply; Associated press; https://blog.ap.org/
announcements/automated-­earnings-­stories-­multiply
3. ChatGPT-­maker OpenAI signs deal with AP to license news stories; AP News;
https://apnews.com/article/openai-­c hatgpt-­a ssociated-­press-­ap-­f 86f84c5
bcc2f3b98074b38521f5f75a
4. BuzzFeed Says AI Will ‘Replace Majority of Static Content’; Futurism; https://
futurism.com/buzzfeed-­ai-­replace-­content
5. Google Tests AI Tool That Is Able to Write News Articles; New York Times;
https://www.nytimes.com/2023/07/19/business/google-­artificial-­intelligence-­
news-­articles.html
6. Bloomberg Develops Generative AI Model Trained On Financial Data; PYMNTS;
https://www.pymnts.com/news/artificial-­i ntelligence/2023/bloomberg-­
develops-­generative-­ai-­model-­trained-­financial-­data/
7. News Corp using AI to produce 3,000 Australian local news articles a week;
The Guardian; https://www.theguardian.com/media/2023/aug/01/news-­corp-­
ai-­chat-­gpt-­stories
8. MSN Fired Its Human Journalists and Replaced Them with AI That Publishes
Fake News About Mermaids and Bigfoot; Futurism; https://futurism.com/msn-­
is-­publishing-­more-­fake-­news

93
 GENERATIVE AI IN PRACTICE

9. Generative AI and Technology Innovation at Wimbledon 2023; Bernard

Marr; https://bernardmarr.com/generative-­ai-­and-­technology-­innovation-­at-­
wimbledon-­2023/
10. How Artificial Intelligence, Data and Analytics Are Transforming Formula
One in 2023; Forbes; https://www.forbes.com/sites/bernardmarr/2023/07/10/
how-­artificial-­i ntelligence-­d ata-­and-­analytics-­are-­t ransforming-­formula-­
one-­in-­2023/
11. Fox Sports Taps Google Cloud For Generative AI Capabilities; Variety;
https://variety.com/2023/digital/news/fox-­s ports-­g enerative-­a i-­g oogle-­
cloud-­1235706964/
12. Globant and LaLiga Tech to Pilot Generative AI Applications to Reinvent Sports
Tactics and Broadcasting; Globant; https://www.globant.com/news/globant-­
laliga-­tech-­generative-­ai-­applications-­sports
13. An AI-­Written Novella Almost Won a Literary Prize; Smithsonian Magazine;
https://www.smithsonianmag.com/smart-­news/ai-­w ritten-­novella-­a lmost-­
won-­literary-­prize-­180958577/
14. Books Written by Artificial Intelligence: A List; All Good Great; https://
allgoodgreat.com/list-­of-­books-­written-­by-­artificial-­intelligence/
15. Sci-­fi author writes over 100 books in 9 months using AI including ChatGPT
and Midjourney; Times of India; https://timesofindia.indiatimes.com/life-­
style/books/features/tim-­boucher-­sci-­fi-­author-­writes-­over-­100-­books-­in-­9-­
months-­using-­ai-­including-­chatgpt-­and-­midjourney/articleshow/100528872.
cms?from=mdr
16. How to access thousands of free audiobooks, thanks to Microsoft AI and
Project Gutenberg; ZDNet; https://www.zdnet.com/article/heres-­how-­to-­access-­
thousands-­of-­free-­audiobooks-­thanks-­to-­microsoft-­ai-­and-­project-­gutenberg/
17. Death of the narrator? Apple unveils suite of AI-­ voiced audiobooks; The
­G uardian; https://www.theguardian.com/technology/2023/jan/04/apple-­
artificial-­intelligence-­ai-­audiobooks
18. Introducing Voice Translation for podcasters; Spotify; https://newsroom.
spotify.com/2023-­09-­25/ai-­voice-­translation-­pilot-­lex-­fridman-­dax-­shepard-
­steven-­bartlett/
19. Robin Williams’ daughter says she finds AI recreations of her father’s voice ‘per-
sonally disturbing’; Insider; https://www.businessinsider.com/robin-­williams-­
daughter-­zelda-­disturbing-­ai-­recreate-­voice-­2023-­10?r=US&IR=T
20. Everything Everywhere All At Once: How AI is Revolutionizing Filmmaking;
Medium; https://medium.com/@TheTechTrailblazer/everything-­everywhere-­
all-­at-­once-­how-­ai-­is-­revolutionizing-­filmmaking-­173bf19d32be
21. How AI will augment human creativity in film production; Variety; https://­
variety.com/vip/how-­artificial-­intelligence-­will-­augment-­human-­creatives-­in-­
film-­and-­video-­production-­1235672659/
22. Generative AI is bringing the biggest disruption to film making in 100 years;
Tech.eu; https://tech.eu/2023/01/23/flawless-­brings/

94
 A NEW DAWN IN MEDIA AND ENTERTAINMENT

23. Sir Paul McCartney says artificial intelligence has enabled a ‘final’ Beatles song;
BBC News; https://www.bbc.com/news/entertainment-­arts-­65881813
24. Virtual Influencer Noonnoouri Lands Record Deal; Forbes; https://www.forbes
.com/sites/bernardmarr/2023/09/05/virtual-­i nfluencer-­n oonoouri-­l ands-­
record-­deal-­is-­she-­the-­future-­of-­music/
25. Grimes Launched a Platform to Help You Make AI Songs With Her Voice.
Here’s What It Sounds Like; Gozmodo; https://gizmodo.com/grimes-­elon-­musk-­
openai-­ai-­music-­elf-­tech-­1850409972
26. Meet your DJ; Spotify; https://newsroom.spotify.com/2023-­02-­22/spotify-­
debuts-­a-­new-­ai-­dj-­right-­in-­your-­pocket/
27. Spotify spotted developing AI-­generated playlists created with prompts; TechCrunch;
https://techcrunch.com/2023/10/02/spotify-­spotted-­developing-­ai-­generated-­
playlists-­created-­with-­prompts/
28. Spotify removed thousands of AI-­generated songs; Mashable; https://mashable.
com/article/spotify-­ai-­crackdown
29. Open letter: Artists Using Generative AI Demand Seat At Table From US
Congress; Creative Common; https://creativecommons.org/about/policy-­
advocacy-­copyright-­reform/open-­letter-­artists-­using-­generative-­ai-­demand-­
seat-­at-­table-­from-­us-­congress/
30. A portrait created by AI just sold for $432,000. But is it really art?; The Guardian;
https://www.theguardian.com/artanddesign/shortcuts/2018/oct/26/call-­that-­
art-­can-­a-­computer-­be-­a-­painter
31. New Gucci NFTs combine fashion and art using generative AI; Vogue Business;
https://www.voguebusiness.com/technology/new-­gucci-­nfts-­combine-­fashion-­
and-­art-­using-­generative-­ai
32. China’s AI Landscape: Baidu’s Generative AI Innovations in Art and Search; Forbes;
https://www.forbes.com/sites/bernardmarr/2023/09/27/chinas-­ai-­landscape-
­baidus-­generative-­ai-­innovations-­in-­art-­and-­search/

95
 7
ADVERTISING AND
MARKETING: BRIDGING
CREATIVITY WITH AI

When exploring GenAI use cases and appetite among sales and marketing
teams, McKinsey found that 90% of commercial leaders expect to deploy
GenAI solutions “often” over the next two years.1 GenAI is expected to rev-
olutionize the word of advertising and marketing, with the ability to auto-­
generate and personalize content, even in real time. And as we’ll see in this
chapter, GenAI solutions will enable businesses of all sizes to create beautiful,
professional ads and marketing campaigns.

Let’s explore how brands will entice customers in the era of GenAI. Even if
you don’t work in this field, the way in which organizations advertise to you
is about to change, so read on.

Overview: How Can Generative AI Be Applied

in Advertising and Marketing?
GenAI can be used to create professional-­looking content that is not just
engaging for audiences, but also tailored to their specific needs.

97
 GENERATIVE AI IN PRACTICE

The advantages of GenAI in marketing and advertising

One of the biggest benefits is the ability to create personalized content. After
all, GenAI excels at analyzing data, and identifying patterns in the data – in
this case, patterns that shed light on consumer behavior and preferences. It
can then create imaginative content that is highly targeted to the desired audi-
ence, for example, in the form of targeted social media campaigns or person-
alized emails.

Another major benefit comes down to time and money, since GenAI can
automate many of the processes involved in creating marketing and adver-
tising materials. This allows marketers to focus their time on more strategic
efforts – and focus their budget on high-­impact activities.

Plus, GenAI allows even smaller teams to increase innovation and generate
inspirational ads and marketing materials.

So what can brands create with GenAI?

We’ll cover plenty of real-­world use cases in this chapter, but as a general
guide, GenAI can be used in marketing and advertising to:

• Generate fresh ideas that boost brainstorming sessions.

• Generate beautiful images – from images of products to imaginative
visuals for aspirational ads. It can also be used to create branding col-
lateral, including graphics and logos.
• Generate text for a variety of purposes, including emails, ad copy, blog
posts, social media posts, product descriptions, and even scripts for ads.
• Generate video, including video ads for social media (or broadcasting),
and product demos.
• Generate music to accompany videos.
• Plus, GenAI can be used as a sentiment analysis tool, analyzing data
for sentiment (e.g., looking at social media mentions to see whether
customer mentions are positive, negative, or neutral).

98
 ADVERTISING AND MARKETING: BRIDGING CREATIVITY WITH AI

Handle with care

We’ve already talked about the potential pitfalls and concerns around GenAI
(see Chapter 4), so professionals must take care to implement GenAI in a safe,
ethical way. We’ll talk more about successful implementation in Chapter 18,
but here are a few tips to get you thinking:

• It’s vital you balance automation with the human touch. Remember,
GenAI mimics human creativity but it is not human. There’s no sub-
stitute for human sensitivity, empathy, and cultural awareness. Plus,
you will know your brand, its values, and ethics better than a computer
model. So always add human knowledge and experience to ensure you
get the best out of GenAI.
• Practice the highest levels of caution around personal data, copyright
restrictions, and other legal issues. And do be aware of the potential for
bias to creep in.
• And as always, be transparent with stakeholders about how you’re
using GenAI.

Using Generative AI to Create Ads and

Marketing Materials
With the ability to generate text, images, video, and music, it’s clear that
GenAI has a lot of potential to create ads and marketing materials – with far
less work and expense than traditional methods.

Automating aspects of ad creation

At the Adobe 2023 EMEA Summit, I saw a demonstration of Adobe’s Firefly
GenAI package, and let me tell you, it blew me away with its capabilities.
Firefly is a natural language image generation engine that takes a simple text
prompt and uses it to create pictures, designs, or any kind of artwork. Fire-
fly has been integrated into the Adobe Express content generation platform,
which allows non-­professional designers to create high-­quality images for
social media and promotional content. (Firefly is also being integrated into
other Creative Clouds applications like Photoshop and Illustrator.)2

99
 GENERATIVE AI IN PRACTICE

In the demonstration, we were shown a photo of a beautiful wild island in

what looked like Canada. Then the demonstrator prompted the system to add
a hiker with a branded backpack recreating on the island . . . which it did with
seamless results. How about adding a canoe in the water, to show how the
hiker got there? No problem, done (and with totally realistic shading, I might
add). Then, the demonstrator asked for the picture to be changed from portrait
orientation to landscape. But to do that, the system had to fill in the “blanks”
at the sides of the image – again, something it did seamlessly. Then, finally,
they added in brand text. It was a stunning demonstration of the power of
GenAI, and how it will allow businesses of all shapes and sizes to create beau-
tiful advertising and marketing content with ease. (For the record, Firefly is
trained exclusively on licensed or public domain images, and the company has
said it will indemnify customers against copyright claims, as per Chapter 4.)

Examples of GenAI ads

It’s easy to see why brands would choose AI-­generated ads over traditional
ads. It’s significantly cheaper and easier, allowing brands to increase output
while spending less; plus, it allows for unlimited creativity, free from real-­
world constraints. So let’s look at a few real-­world examples.

Big-­name brands like Oreo and Cadbury have used OpenAI’s DALL-­E 2 to
create ads (in Cadbury’s case, creating an ad for the Indian market featuring
an AI version of Bollywood star Shah Rukh Khan).3 Fashion brands Nike and
Tommy Hilfiger have used Midjourney to create ad images featuring their
products.4 And Coca-­Cola deployed ChatGPT and DALL-­E 2 in a stunning
ad campaign called “Masterpiece.” The ad brings to life some of the world’s
most famous works of art, seamlessly integrating AI-­augmented animation
with live action. I really like this example because it shows that GenAI can be
used not just to churn out ads on the cheap, but also to create significant wow
factor. Coca-­Cola is so invested in GenAI as a transformational tool that it has
appointed a Head of Generative AI.5

Meanwhile, Heinz fed DALL-­E 2 prompts such as “ketchup in outer space”

(and lots more fun ideas), posting the resulting images on their social media
accounts. Heinz also encouraged consumers to create and share their own AI

100
 ADVERTISING AND MARKETING: BRIDGING CREATIVITY WITH AI

ketchup images, helping the campaign to go viral.6 This taps into the huge
move toward personalization in advertising and marketing, but we’ll talk
more about that later in the chapter.

New tools to help brands leverage GenAI

In Chapter 3, I mentioned that Meta is creating tools that allow businesses
to automatically make multiple versions of Facebook ads, featuring differ-
ent text and images aimed at different audiences.7 But Meta isn’t the only
company promising to help businesses create GenAI ads and marketing copy
more easily.

In 2023, Salesforce announced Einstein GPT, which it bills as the “world’s first
generative AI for customer relationship management.”8 The tool uses GenAI
to create personalized content across Salesforce cloud, meaning Salesforce
customers can connect their data and use natural language prompts to gener-
ate content that adapts to changing customer information and needs “in real
time.” (Again, more on personalization coming up.)

Another example comes from Aprimo, providers of marketing automation

software. Aprimo has created an “embedded generative AI assistant” using
Azure OpenAI and ChatGPT that can be used to, among other things, create
engaging and personalized content that’s trained on brand voice and tone.9
Another service built on Microsoft Azure is Typeface, which uses GenAI to
“create engaging, on-­brand marketing messages in seconds.”10 The idea is you
feed Typeface your style guidelines, images, and product details, and you can
then start generating suggested images and text.

Elsewhere, LinkedIn has launched a new feature that allows B2B marketers to
use GenAI. Called “Accelerate,” the tool aims to speed up ad campaign setup
and automate ad optimization. LinkedIn claims it will help B2B marketers
create tailored creative content in as little as five minutes.11

A similar movement is beginning to happen in TV advertising. Fox TV has

partnered with AI video generation specialists Waymark to enable GenAI
commercials across all Fox stations, including local stations.12 This will make

101
 GENERATIVE AI IN PRACTICE

TV advertising much more accessible for small and local businesses – ­allowing
them to easily create high-­quality TV ads with limited time and budget.

Personalization of Content with GenAI

Personalization is a recurring theme throughout this book, and it’s especially
relevant in the field of marketing and advertising.

Personalization at scale
Brands know that we consumers crave that feeling of uniqueness and exclu-
sivity – that we want to feel like we have a personal relationship with the
brands we love, and that our support matters. But, achieving this effect on
a mass scale is difficult and expensive. Or at least, it has been up until now.
Because GenAI makes it possible for brands to deliver the uniqueness and
personal touch that was previously the domain of high-­end brands only.

Which is why more and more GenAI tools are including functionality around
personalization. Adobe’s Experience Cloud is a collection of tools includ-
ing the Adobe Real Time Customer Data Platform, Journey Optimizer, and
Adobe Analytics. These tools use AI to deliver insights on customer jour-
neys, behaviors, and engagement. Information which, when combined with
GenAI, can be used to automatically create hyper-­personalized content.

The key, of course, is achieving that bespoke feeling without creeping people
out or giving them the impression that their privacy has been invaded. Which
is another thing that brands will have to consider carefully when deploying
GenAI in highly personalized campaigns.

Examples of personalized experiences

Now let’s look at a few examples. Netflix uses AI to create highly targeted
thumbnails for its streaming content to increase the likelihood that customers
will click and watch. The streaming giant has thousands of frames from each
movie or TV show that can be used to automatically generate thumbnails – but

102
 ADVERTISING AND MARKETING: BRIDGING CREATIVITY WITH AI

did you know that the thumbnails you see could be entirely different to the
thumbnails that I see? Netflix uses AI to analyze and rank images according
to which have the highest likelihood of getting you to click – based on what
others who are deemed similar to you have clicked on.13

What about personalized advertising campaigns? The Barbie Movie delivered

a masterclass in movie promotion, seemingly taking over the internet for sev-
eral weeks in 2023. One of the campaign’s biggest hits was the Barbie Selfie
Generator,14 a website that let fans create personalized versions of the movie’s
poster featuring themselves (or whomever they wanted). Needless to say, it
went viral. Which goes to show consumers’ appetite for digital, interactive
experiences that can be easily personalized (and, ideally, shared with friends).

Combine personalization with the ability to create AI versions of stars, and

you have the potential to deliver some very fun experiences for consum-
ers – rather like the Heinz example from earlier in the chapter, but featuring
real-­world celebrities. Superstar footballer Lionel Messi signed a deal with
PepsiCo allowing the brand to use a deepfake version of himself to promote
Lays chips.15 As part of the Lays Messi Message campaign, users can cre-
ate personalized messages from Messi to themselves, in English, Spanish,
­Portuguese, and Turkish. And it’s all possible thanks to GenAI.

However, there’s a dark side to this ability to deepfake our favorite celebrities.
Just as I was writing this chapter, Tom Hanks posted on social media to warn
fans about a video of him appearing to promote a dental plan.16 The video was
fake and Hanks had nothing to do with it, which is pretty scary.

AI Influencers and Models

In Chapter 6, I mentioned Noonoouri, the AI-­generated influencer who signed
a record deal with Warner. Let’s explore virtual influencers and models in a
­little more detail, and see how they are becoming more important for brands.

A virtual influencer is a digital avatar that offers total customization and 24/7
availability to fans, typically via social media. They can take the shape of a
realistic-­looking human, a cartoon human, or a non-­human creation. This isn’t a

103
 GENERATIVE AI IN PRACTICE

new phenomenon, by the way – the first virtual popstar was a Japanese ­character
called “Kyoko Date,” created in the 1990s. What’s new here is the intersection
of social media (which allows these “personalities” to reach a vast global audi-
ence), GenAI (which brings unlimited creative potential), and the concept of the
metaverse (where the real and digital worlds increasingly merge together).

Some virtual influencers, such as Lil Miquela and Lu do Magalu, have ­millions
of followers. Yet, for brands looking to partner with an influencer, virtual
influencers often prove distinctly more affordable than human influencers.
But are brands really using virtual creations? It appears so. Fashion brand
Prada has its own virtual “muse” called “Candy,” unveiled in 2021 to relaunch
the Prada Candy fragrance.17 Virtual influencer Lil Miquela has collaborated
with several brands, including Calvin Klein, and is reportedly worth as much
as US$125 million.18 And we’re not just talking about virtual social media
stars, but also virtual fashion and beauty models. Shudu Gram was labeled
the “world’s first avatar supermodel,” and has collaborated with brands such
as Fenty.19 Levi’s has also announced that it will be experimenting with AI-­
generated models to enhance diversity, collaborating with digital fashion stu-
dio Lalaland.ai to create hyper-­realistic models.20

When you team up with a virtual influencer or model, you can create endless
campaigns and images via GenAI, without needing to hire a studio, photog-
rapher, etc. These AI creations don’t age (unless you want them to). They
can represent a multitude of skin tones and body types. They can be restyled
instantly in line with new trends, without the need for expensive reshoots.
And they’ll never get bogged down in a drugs or sex scandal. As such, virtual
influencers and models have the potential to transform what it means to cre-
ate, market, and consume content online.

However, there are issues. Transparency is one – meaning, as these influenc-

ers become more lifelike, how should their creators make it clear that these are
not real beings? There’s also a danger that virtual influencers and models may
perpetuate unrealistic beauty and body type standards (although, the same
can be said of the majority of ad campaigns). Then there are issues of who
exactly is behind these digital creations. In the case of Shudu, a young black
woman, her creator is a white male – prompting some critics to describe his
creation as appropriation, expropriation, or at best jumping on the trend for

104
 ADVERTISING AND MARKETING: BRIDGING CREATIVITY WITH AI

darker-­skinned models.21 And finally, there’s the argument that these ­virtual
influencers and models lack the human connection that you get with a real,
living, breathing, imperfect human being. Can they really offer the kind of
connection with audiences that brands crave? Time will tell.

But whether you’re intrigued or appalled by the idea of virtual influencers and
models, there’s no doubt that brands are interested in experimenting with this
new approach.

Key Takeaways
In this chapter, we’ve learned:

• GenAI can be used to help brands generate text, visuals, video, and
music for high-­quality advertising and marketing campaigns – far
cheaper and quicker than through traditional methods. As such, GenAI
will allow businesses of all sizes to create beautiful images, engaging
copy, and more.
• GenAI can help businesses come up with ideas, and automate aspects
of the creative process – for example, by creating images through natu-
ral text prompts. Several high-­profile brands have already embraced
GenAI for ads.
• New tools are emerging that enable brands to leverage GenAI possibili-
ties, including on platforms like Facebook and LinkedIn.
• The ability to personalize ads and marketing materials on a mass scale
is a major advantage of GenAI – from text and images that are tailored
to audience preferences, to creating interactive deepfake video experi-
ences featuring celebrities.
• Brands are also choosing to work with virtual AI influencers and
­models – creating images, campaigns, and videos without the need to
hire human models and influencers (and photographers, studios, etc.).

Marketing and advertising are all about engaging customers with thoughtful,
inspiring, and compelling content. But how else are brands using GenAI to
engage with customers and provide a better service? Turn the page to find out. . .

105
 GENERATIVE AI IN PRACTICE

Notes
1. AI-­powered marketing and sales reach new heights with generative AI;
McKinsey & Company; https://www.mckinsey.com/capabilities/growth-­
marketing-­and-­sales/our-­insights/ai-­powered-­marketing-­and-­sales-­reach-­new-­
heights-­with-­generative-­ai
2. Unleashing Creativity with AI: Adobe’s Trailblazing Generative Tools at EMEA
Summit 2023; Bernard Marr; https://bernardmarr.com/unleashing-­creativity-­
with-­ai-­adobes-­trailblazing-­generative-­tools-­at-­emea-­summit-­2023/
3. Big companies use AI-­generated ads because they’re cheap; The Verge; https://
www.theverge.com/2023/8/18/23837273/generative-­ai-­a dvertising-­oreos-­
cadbury-­watermarking
4. Fashion brands embrace Midjourney for captivating imagery; Design & Build
Co; https://designandbuild.co/insights/fashion-­brands-­embrace-­midjourney-­
for-­captivating-­imagery
5. The Amazing Ways Coca-­Cola Uses Generative AI in Art and Advertising;
Forbes; https://www.forbes.com/sites/bernardmarr/2023/09/08/the-­amazing-­
ways-­coca-­cola-­uses-­generative-­ai-­in-­art-­and-­advertising/
6. Top AI-­Generated Advertising Campaigns from Famous Brands; Digital Agency
Network; https://digitalagencynetwork.com/top-­ai-­generated-­a dvertising-­
campaigns-­from-­famous-­brands/
7. 5 Amazing Ways How Meta (Facebook) Is Using Generative AI; Bernard Marr;
https://bernardmarr.com/5-­a mazing-­w ays-­h ow-­m eta-­f acebook-­i s-­u sing-­
generative-­ai/
8. Salesforce Announces Einstein GPT, the World’s First Generative AI for CRM;
Salesforce; https://www.salesforce.com/uk/news/press-­releases/2023/03/07/
einstein-­generative-­ai/
9. Aprimo launches first ChatGPT within Azure OpenAI DAM integration, giving
marketers enhanced, scalable automation; Microsoft; https://customers.microsoft
.com/en-­us/story/1647646542947823380-­aprimo-­professional-­services-­azure-­
openai-­service
10. Typeface uses Azure OpenAI Service to create engaging, on-­brand marketing
messages in seconds; Microsoft; https://customers.microsoft.com/en-­us/story/
1637196302736110361-­typeface-­professional-­services-­azure-­openai-­service
11. LinkedIn Introduces Generative AI-­Powered Tool for B2B Marketers; Adweek;
https://www.adweek.com/brand-­marketing/linkedin-­generative-­ai-­powered-­
tool-­b2b-­marketers/
12. Fox Stations to Offer AI Generated Commercials Through New Partnership;
Adweek; https://www.adweek.com/tvspy/fox-­stations-­to-­offer-­ai-­generated-­
commercials-­through-­new-­partnership/250658/
13. How Netflix Uses AI, Data Science And Machine Learning; Medium; https://
becominghuman.ai/how-­netflix-­uses-­ai-­and-­machine-­learning-­a087614630fe
14. Selfie generator; Barbie Selfie AI; https://www.barbieselfie.ai/uk/

106
 ADVERTISING AND MARKETING: BRIDGING CREATIVITY WITH AI

15. Why some celebrities are embracing Artificial Intelligence deepfakes; BBC News;
https://www.bbc.com/news/business-­65995089
16. Tom Hanks Warns of Deepfake Hanks Dental Ad; Voicebot; https://voicebot
.ai/2023/10/02/tom-­hanks-­warns-­of-­deepfake-­hanks-­dental-­ad/
17. Prada Creates Virtual Muse Named Candy; Virtual Humans; https://www
.virtualhumans.org/article/prada-­creates-­first-­virtual-­muse-­candy
18. Meet Miquela: The AI (Artificial Influencer) Who’s Now Worth $125 Million;
Medium; https://medium.com/illumination/meet-­miquela-­the-­a-­i-­artificial-­
influencer-­whos-­now-­worth-­125-­million-­879e4da7baf0
19. Shudu: Fashion’s First Avatar Supermodel?; WWD; https://wwd.com/eye/­
people/shudu-­digital-­fashion-­model-­avatar-­1202683320/
20. Pixel Perfect: The Rise of AI Fashion Models; Forbes; https://www.forbes.com/
sites/bernardmarr/2023/06/07/pixel-­perfect-­the-­rise-­of-­ai-­fashion-­models/
21. Shudu Gram is a white man’s digital perception of real-­life black womanhood;
The New Yorker; https://www.newyorker.com/culture/culture-­desk/shudu-­gram-­
is-­a-­white-­mans-­digital-­projection-­of-­real-­life-­black-­womanhood

107
 8
REINVENTING CUSTOMER
ENGAGEMENT THROUGH
INTELLIGENT SYSTEMS

As we’ve already seen in this book, the customer service function stands to
be revolutionized by GenAI, particularly by text-­and-­speech models. But
­customer service isn’t the only aspect of customer engagement that can ben-
efit from GenAI; through new personalized offerings, preventive interven-
tions, and intelligent products and services, GenAI will transform how brands
interact with their customers. In short, GenAI offers unparalleled opportuni-
ties for personalization, responsiveness, and increased customer satisfaction.

Let’s explore how GenAI can bring brands and their customers closer together.

Generative AI in Customer Service

In Chapter 3, we saw how UK energy supplier, Octopus Energy, has deployed
ChatGPT in customer service – to the extent that the bot now handles 44%
of customer enquiries, does the work of 250 people, and receives higher cus-
tomer satisfaction ratings than human customer service agents.1 It’s a prime
example of how contact centers will increasingly incorporate GenAI chat and
voice tools to deal with straightforward, easily repeatable tasks.

109
 GENERATIVE AI IN PRACTICE

The capabilities of GenAI

You may think customers don’t want to talk to bots, but the Octopus Energy
example shows that GenAI tools can deliver impressive results and bring a new
level of responsiveness to customer service bots. Traditional AI offerings (like
some of the not-­very-­intelligent chatbots you might have interacted with) rely
on rules-­based systems to provide predetermined responses to questions. And
when they come up against a query that they don’t recognize, or doesn’t follow
defined rules, they’re stuck. And even when they do give a helpful answer, the
language is typically pretty stiff. GenAI, on the other hand, can understand
even complex questions and answer in a more natural, conversational way.
And, of course, these tools give customers 24/7 access to support, via multiple
channels (such as phone, online chat, and social media messaging).

But answering customer queries isn’t the only way GenAI can add value
in customer service. Some of the other tasks that GenAI can do or assist
with include:

• Giving customers personalized recommendations, based on the cus-

tomer’s data and previous interactions, which further helps to enhance
the customer experience. (More on personalization coming up later.)
• Providing conversational search functions for, say, FAQs online.
GenAI can take natural language prompts like “where’s my package”
or “I hate the sweater I ordered” and either direct the customer to the
correct FAQ response, or deliver a tailored response. And don’t forget
this can be done in multiple languages.
• Optimizing data to support customer service operations. GenAI can
handle vast amounts of data and turn that information into actionable
insights – insights such as “Which are our most common complaints?”
It can also track and categorize customer trends with ease.
• Supporting human customer service agents. GenAI can help human
agents be more productive. For example, it can automatically generate
responses to common queries, provide summaries of previous com-
plaints and resolutions that agents can use in conversations, and gener-
ate product recommendations.

110
 REINVENTING CUSTOMER ENGAGEMENT THROUGH INTELLIGENT SYSTEMS

In this way, GenAI can help improve the employee experience as well as the
customer experience. It can support the work that human agents do, and free
them up to focus on more complex customer interactions where they can add
the most value. It’s no wonder customer service has become CEOs’ number-
one GenAI priority, according to the IBM Institute for Business Value, with
85% of execs saying GenAI will be interacting directly with their customers
within the next two years.2

But how to go about implementing GenAI without alienating employees? In their

book Human + Machine: Reimagining Work in the Age of AI, Paul R. Daugherty
and H. James Wilson argue that GenAI will enrich, not replace, customer service
jobs. They recommend organizations break jobs down into underlying tasks,
then explore how GenAI will affect each task. For customer service, they found
that most customer service work could be broken down into 13 tasks. Four of
those tasks remained unchanged by GenAI, four could be fully automated by
GenAI, and five could be augmented by GenAI. Even better, five new high-­value
tasks emerged thanks to GenAI taking on the easily automated tasks.

As with all aspects of GenAI, transparency is a must – by which I mean mak-

ing it clear when customers are talking to a bot. Read more about successful
implementation in Chapter 18.

Examples from customer service

Now let’s explore some real-­world examples of GenAI tools and companies
who are using GenAI to enhance customer service. One example of GenAI
designed for customer service is Zendesk’s Expanding Agent Replies solu-
tion, which is like predictive text on steroids. Agents can type the bare bones
of a response to a customer, and then the tool fleshes it out for them. Or
there’s Google Cloud’s Generative FAQ for CCAI Insights, which lets contact
centers upload (redacted) transcripts of calls and unearth customer FAQs.
Similarly, Twilio’s CustomerAI solution can generate articles to plug gaps in
knowledge – articles which can be published for customers or digested by
agents to better improve outcomes. Or there’s Sprinklr’s call note automation
tool, which automatically notes crucial info as the customer talks – allowing
the agent to focus on active listening.

111
 GENERATIVE AI IN PRACTICE

JetBlue has partnered with technology company ASAPP to implement GenAI,

with impressive results. JetBlue’s contact center has saved an average of 280
seconds per chat, freeing up 73 000 hours of agent time in just one quarter.3
That’s a whole lot of hours to devote to customers with more in-­depth prob-
lems. Another airline, Air India, has said it’s deploying OpenAI’s ChatGPT as
part of a huge system modernization. Chatbots have reportedly been used to
modernize the airline’s website and mobile apps, digital marketing, customer
notifications system, customer service portal (including real-­time support
request tracking), and contact center operations.4

Meanwhile, British telecom company BT Group used GenAI capabilities in

Microsoft Power Platform to create a digital assistant called “Aimee” to help
customers with a variety of queries. BT says the in-­app messaging assistant
has answered approximately 60% of customer queries with great customer
satisfaction (Net Prompter Score of 60).5

In India, digital payments company PhonePe uses the Freddy AI bot devel-
oped by Freshworks to automate answers for simple queries such as “What’s my
­balance?” – offering a personalized self-­service experience.6 Which brings us to. . .

A New Level of Personalization

Clearly, GenAI can be used to deliver a more personalized customer ser-
vice experience, but it can also be used to personalize all sorts of customer
interactions. In the future, most systems and services across most industries
will incorporate some element of AI in order to better understand customer
­preferences and customize the service accordingly. (This will happen in eve-
ryday home items, too, but we’ll talk more about that later.)

Personalization everywhere
Take the automotive industry as an example. GenAI-­powered in-­car systems
could generate customized driving experiences. So, based on my history and
preferences, my car could potentially adjust the cabin ambience, suggest new
music playlists or podcasts, and automatically select driving modes according

112
 REINVENTING CUSTOMER ENGAGEMENT THROUGH INTELLIGENT SYSTEMS

to the terrain and weather conditions – all of which will enhance the driving
experience.

In the hospitality sector, AI-­enhanced hotel rooms could offer guests an envi-
ronment that anticipates their preferences for lighting and temperature. GenAI
could also be used to generate recommendations for local activities and dining
options, based on the guest’s past choices. Similarly, travel agencies could use
GenAI tools like ChatGPT to curate bespoke travel itineraries for individuals
based on their travel history, hobbies, and even social media activity. The itin-
erary for a food enthusiast, for example, would be entirely different to that of a
nature lover – all automatically generated based on customer data.

In entertainment, streaming platforms could harness GenAI to curate per-

sonalized movie or TV recommendations, not just based on what they’ve
watched in the past but by analyzing plot preferences, character archetypes,
or cinematographic styles favored by the viewer.

And in the realm of fitness, wearable technology (such as wristbands) inte-

grated with GenAI can craft personalized workout routines or training regi-
mens. For example, if a user has been running frequently, the AI might suggest
cross-­training exercises to balance muscle development, or even curate a list
of local races or events they might be interested in.

These are just a few of the ways industries may offer increasingly personalized
solutions, powered by GenAI.

PGA and personalized fan engagement

We’ve already seen several real-­world examples of personalization through-
out this book (such as Stitch Fix’s ability to provide personalized outfit recom-
mendations, Chapter 3) – and we’ll see many more in the remaining chapters.
Personalization is certainly a recurring theme within GenAI. But here’s
another example to highlight the possibilities.

The PGA Tour has partnered with Amazon Web Services (AWS) to pro-
vide a better, more personalized viewing experience for golf fans, powered

113
 GENERATIVE AI IN PRACTICE

by GenAI. The idea is that AWS machine-­learning technology will be used to

leverage hundreds of hours of raw tournament footage and turn the content
into new digital offerings that give fans their own, unique viewing experience.
Among the plans is a streaming platform called “Every Shot Live” that will
give fans live access to every shot played by every player – comprising more
than 32 000 shots by 144 golfers.7 Essentially, fans will be able to follow any
player they like in real time, and experience cool features like changing the
camera angle or viewing game stats on-­demand.

Delivering More Proactive Interventions through

Predictive AI
Personalization also enables more proactive interventions – meaning, by
understanding how a customer uses your product or service, you can pre-­
empt what they want next and deliver it seamlessly. This is already happen-
ing across many industries – predictive maintenance for machinery being a
prime example – but the key takeaway here is that GenAI will give predictive
capabilities an extra boost, thereby enabling more proactive interventions
that delight and impress customers.

What might this look like in future?

Utilities companies could certainly deploy AI to proactively address user con-
cerns. Say, for example, a consumer’s usage pattern suddenly deviates from
the norm, a GenAI system could proactively reach out to check in with the
customer or provide troubleshooting solutions – all before the consumer even
notices there’s an issue.

It’s easy to imagine similar options across all sorts of industries. In the
­telecommunications sector, for example, AI could analyze call quality, data
consumption, and service interruptions to pre-­emptively offer better solu-
tions, or even create personalized service packages that precisely match a
user’s consumption patterns.

In transportation, commuters could definitely benefit from a more proactive

service. If someone frequently travels a certain route, which is expected to

114
 REINVENTING CUSTOMER ENGAGEMENT THROUGH INTELLIGENT SYSTEMS

have delays or construction, the public transport network could use GenAI to
proactively suggest alternative routes or methods of transportation.

Predictive fintech solutions

The financial and banking sector is another area that could benefit enor-
mously from GenAI-­led predictive solutions. Indeed, this is already hap-
pening to a large extent, with AI-­driven automatic fraud detection and risk
assessment. But imagine how GenAI could take this to a whole new level
with, say, personalized financial advice and planning. One study found that
using GenAI to deliver personalized financial advice could save customers on
average US$1200 a year.8 JPMorgan Chase is one organization using GenAI
to create personalized financial plans, tailored to customers’ individual needs
and financial goals.9 Elsewhere, Visa has launched a US$100 million initiative
to invest in GenAI solutions that enhance commerce and payments – likely
focusing on fraud prevention, personalized marketing campaigns for busi-
ness, and potentially even developing new payment methods.10

Incorporating Generative AI into Products

I’ve already said that I anticipate cars becoming more intelligent and offering
a more personalized experience powered by AI. But it won’t stop there. . .

How GenAI could be built into everyday devices

We’ll likely see ovens, washing machines, and other everyday home items
with built-­in GenAI. Indeed, AI has been creeping into our homes for years
(think smart lightbulbs and Alexa) – but thanks to GenAI, these interactions
will become even more human and more personal. We could, for example,
engage in natural language conversations with home appliances. Yes, you
could potentially ask your washing machine questions, such as whether it’s
safe to wash a beloved item of clothing on a certain setting. Or you could say
to your fridge, “Hey, when am I going to run out of milk?” and it’ll tell you.

Integrating GenAI into everyday products could lead to a new era of smart
appliances that are not only more adaptive to our needs, but also more inter-
active and engaging. Engaging ovens? Why not!

115
 GENERATIVE AI IN PRACTICE

Here’s what this might look like in the future:

More personalized user experiences

• Thanks to ChatGPT’s new capabilities to “see,” you can already
take a picture of the contents of your fridge and ask it what to cook
for dinner. But in the future, you could ask your fridge directly, as
appliances begin to incorporate GenAI assistants.
• Appliances will increasingly be able to adapt to our needs. A wash-
ing machine, for example, could adapt to the types of clothes you
generally wash (e.g., heavy-­duty work clothes that get really dirty)
and create customized cycles for optimal fabric care and cleanliness.
• Similarly, your oven could learn your cooking style and preferences
over time, suggesting optimal temperature settings and cooking
times for different dishes, and even new recipe suggestions.
Interactive problem-­solving
• If you have a problem with an appliance, the GenAI could gener-
ate step-­by-­step troubleshooting guides, tailored to the exact model
and its current conditions. No need to call the contact center or
consult a lengthy user manual.
Predictive maintenance
• Instead of waiting for a part to fail, GenAI could predict when a
component is likely to need replacement or maintenance, and send
you a pre-­emptive alert. It could potentially even order the neces-
sary parts or schedule a service visit ahead of time.
Enhanced customer service
• If you do need to contact customer service, the AI could provide the
agent with a detailed history of the appliance’s usage patterns, any
anomalies, and prior troubleshooting steps – all of which is helpful
context that enables more efficient problem-­solving.
• GenAI could also assist in creating dynamic manuals or interactive
tutorials, providing users with information when they need it.

116
 REINVENTING CUSTOMER ENGAGEMENT THROUGH INTELLIGENT SYSTEMS

Current examples of GenAI in devices

If you think my predictions are a bit wacky, perhaps these real-­world exam-
ples will convince you...

Samsung is reportedly set to incorporate “neural processing units” into all

new home appliances, including smart TVs and refrigerators, from 2024 –
enabling smart, 24/7 GenAI tools that can assist users.11 It’s not yet clear
exactly what features will be offered, but a smart oven that can warn you when
something’s burning is one possible application.

German appliance manufacturer Miele is also incorporating GenAI into its

Smart Food ID cooking assistance system – initially for large ovens with inte-
grated cameras.12 The camera takes pictures of the ingredients, and the AI
interprets the picture and suggests the cooking mode automatically. The user
just has to confirm, and the oven takes care of the rest. AI will also be used by
Miele for appliance diagnostics, to help customers resolve issues when mal-
functions occur. One example is a washing machine creating too much foam
because the user has added too much detergent – in which case, the AI would
recommend the correct amount of detergent for the load amount.

LG is also incorporating AI to help customers use appliances in a better way.

LG’s AI can notify users when an issue occurs, such as a decline in refrigerator
cooling, before it gets any worse. Issues like this aren’t immediately obvious
to the user but could shorten the appliance’s lifespan. The AI can provide
quick-­fix solutions, or schedule an appointment with a service agent. It can
also provide routine maintenance suggestions, like running a clean cycle in
a washing machine every 30 washes.13 (Thus linking back to the ability for
GenAI to predict future issues and provide proactive solutions.)

We’ve already briefly mentioned cars, but let’s explore a couple of examples.
Continental is partnering with Google Cloud to bring GenAI natural lan-
guage technology to cars – basically, enabling drivers to have conversations
with their cars on, say, which route is best or whether the tires need pump-
ing up.14 Meanwhile, Mercedes-­Benz has partnered with Microsoft to bring
in-­car ChatGPT to more than 900 000 vehicles via the “Hey Mercedes” voice

117
 GENERATIVE AI IN PRACTICE

assistant.15 ChatGPT enhances the existing capabilities of Hey Mercedes,

­offering more expansive topics and improved natural language understanding.

There’s no doubt that we’ve been moving toward “AI everything” for quite
some time, but GenAI will accelerate these developments and bring new,
more intuitive AI capabilities to everyday devices. What products will you be
“chatting” to in future? Watch this space. . .

Key Takeaways
In this chapter, we’ve learned:

• In customer service, GenAI can obviously deal with customers directly,

but also carry out or assist with other tasks, such as identifying cus-
tomer service trends, generating responses for human agents, and
helping agents deliver more personalized responses. Rather than com-
pletely replacing human customer service agents, GenAI will take on
the lower-­value tasks, freeing up agents to focus on higher-­value work.
• Personalization is a huge theme in GenAI, and we can expect all organi-
zations to unearth new opportunities to provide personalized solutions
for customers. GenAI can assist with this by understanding the cus-
tomer’s history and preferences, and generating thoughtful suggestions
and recommendations.
• Personalization also enables a new level of proactive responsiveness
for organizations. From identifying potential issues and malfunctions
before they occur, to delivering proactive advice that helps customers
achieve their goals, GenAI can help businesses better anticipate their
customers’ needs.
• And finally, we can expect to see GenAI built into all sorts of everyday
devices and appliances, such as cars, ovens, and washing machines. AI
has been creeping into devices for years, but GenAI will accelerate this
trend and allow for more intuitive, natural language conversations with
everyday items.

118
 REINVENTING CUSTOMER ENGAGEMENT THROUGH INTELLIGENT SYSTEMS

Customer engagement is a topic that applies to any organization across any

industry. Next, we’ll drill into a specific sector to explore GenAI opportuni-
ties. Turn the page to read about GenAI uses in retail.

Notes
1. AI doing the work of over 200 people at Octopus, chief executive says; CityAM;
https://www.cityam.com/ai-­doing-­the-­work-­of-­over-­200-­people-­at-­octopus-­
chief-­executive-­says/
2. From cost center to value creator; IBM Institute for Business Value; https://
www.ibm.com/thought-­leadership/institute-­business-­value/en-­us/report/ceo-­
generative-­ai/customer-­service
3. How Generative AI Is Already Transforming Customer Service; BCG;
https://www.bcg.com/publications/2023/how-­g enerative-­a i-­t ransforms-­
customer-­service
4. ChatGPT will be used by Air India as part of system modernization; Analytics
Insight; https://www.analyticsinsight.net/chatgpt-­will-­be-­used-­by-­air-­india-­as-­
part-­of-­system-­modernization/
5. Strategic Progress report; BT; https://www.bt.com/about/annual-­reports/
2022summary/assets/documents/Strategic_progress_2.pdf
6. Say hello to Freddy AI; Freshworks; https://www.freshworks.com/freddy-­ai/
7. PGA Tour teams up with AWS to improve fan engagement during tournaments;
Computer Weekly; https://www.computerweekly.com/news/252497159/PGA-­
Tour-­teams-­up-­with-­AWS-­to-­improve-­fan-­engagement-­during-­tournaments
8. Exploring the Transformative Potential of Generative AI in Fintech and Bank-
ing; Medium; https://medium.com/@seyhunak/exploring-­the-­transformative-­
potential-­of-­generative-­ai-­in-­fintech-­and-­banking-­c01977c33a64
9. Generative AI for Banking and Financial Services; Medium; https://medium.
com/@jeevan_6696/generative-­ai-­for-­b anking-­f inancial-­s ervices-­t he-­next-­
frontier-­c98d5c42ba30#
10. Visa Invests $100M in Generative AI to Transform the Way We Shop and Pay;
Medium; https://medium.com/@jhakon/visa-­invests-­100m-­in-­generative-­ai-­to-­
transform-­the-­way-­we-­shop-­and-­pay-­4f18dfd9e268
11. Samsung may be adding AI to its home appliances; TechRadar; https://www
.techradar.com/computing/artificial-­intelligence/samsung-­may-­be-­adding-­ai-­
to-­its-­home-­appliances-­im-­so-­ready-­to-­chat-­to-­my-­oven
12. Artificial intelligence for best taste as well as for self-­help error prevention in
case of appliance malfunctions; Miele; https://www.miele.de/en/m/artificial-­
intelligence-­for-­best-­taste-­as-­well-­as-­for-­self-­help-­and-­error-­prevention-­in-­
case-­of-­appliance-­malfunctions-­6669.htm

119
 GENERATIVE AI IN PRACTICE

13. How LG Is Using AI to Change How Consumers User Appliances, Improve

Sustainability and Do Our Laundry Better; Forbes; https://www.forbes.com/
sites/amandalauren/2020/01/20/how-­lgs-­artificial-­intelligence-­is-­changing-­
how-­c onsumers-­u se-­appliances-­i mproving-­sustainability-­and-­d oing-­our-­
laundry-­better/#
14. Google Cloud and Continental to equip generative AI in cars; Techwire Asia;
https://techwireasia.com/2023/09/how-­i s-­g enerative-­a i-­i n-­c ars-­m aking-­
talking-­cars-­real/
15. Mercedes-­Benz takes in-­car voice control to a new level with ChatGPT; Mercedes-­
Benz Group; https://group.mercedes-­b enz.com/innovation/digitalisation/
connectivity/car-­voice-­control-­with-­chatgpt.html

120
 9
RETAIL REIMAGINED:
THE RISE OF VIRTUAL
TRY-­ONS, AI SHOPPING
ASSISTANTS, AND MORE

Incredible GenAI tools like ChatGPT will clearly have an effect on every
industry, and the retail sector is no different. In this chapter, we’ll explore
some of the main ways GenAI can be used in retail – and how it might trans-
form the shopping experience for both customers and retailers. Fashion will
crop up a lot in this chapter but make no mistake, GenAI will impact all kinds
of retailers. We’re talking immersive virtual stores and websites optimized
for GenAI technologies. We’re talking interactive displays in physical stores.
We’re talking hyper-­personalization on a mass scale. And we’re talking opti-
mized behind-­the-­scenes processes. Are you ready to reimagine retail? Let’s
get into it.

Improved Digital Shopping Experiences

From optimized websites to virtual stores in the metaverse, GenAI could
shape the way we shop online.

121
 GENERATIVE AI IN PRACTICE

Virtual storefronts: immersive shopping in the metaverse

With augmented and virtual reality becoming more mainstream, GenAI can
help to create immersive virtual shopping experiences. This ties in with the
concept of the metaverse – immersive virtual worlds where we can shop,
game, work, and more. Forward-­thinking retailers are already building a
presence in the metaverse, creating immersive, interactive and personalized
virtual stores, and brand experiences. Nike is a real leader in the metaverse
space with its part-­game, part-­showroom Nikeland experience on the Roblox
metaverse platform. Nikeland features buildings, arenas, and games, as well as
a virtual showroom complete with digital products.1 Another brand to pitch
up in Roblox is Forever 21, which has created a virtual brand experience that
allows users to “run” their own virtual Forever 21 store.2

GenAI can help to make these virtual experiences more responsive. For exam-
ple, as a user navigates these digital realms, AI can generate personalized store
layouts, product placements, and even virtual sales assistants tailored to the
user’s shopping history and preferences.

Delivering a better website (and app) experience

Outside of the metaverse, GenAI will also enable better, more intuitive experi-
ences on regular websites and apps. Take Expedia Group as an example. Users
can now plan their travel arrangements by having an open-­ended “chat” with the
app, from discussing accommodation to pinpointing the best sights to see. It’s as
close as you can get to the old-­school travel agent experience in app form. Expe-
dia says the feature will also let users know when there are exclusive rewards and
discounts, which they can take advantage of directly via the ChatGPT interface.3

As this chapter shows, we’ll see more and more ecommerce sites incorporate
ChatGPT-­style features into their websites and apps. The technology can be
used to create personalized shopping assistants (more on that coming up),
deliver enhanced customer support, and even be used as a market research
tool, gathering data from customer interactions.

Another way GenAI can prove useful is in summarizing customer reviews.

Ever wanted to check out a product only to find hundreds of reviews offering

122
RETAIL REIMAGINED: THE RISE OF VIRTUAL TRY-­O NS, AI SHOPPING ASSISTANTS, AND MORE

a multitude of (sometimes conflicting) opinions? GenAI can provide ­helpful

summaries that save you wading through all that mess. Online electronics
retailer Newegg has done just that, adding a ChatGPT-­powered feature that
summarizes reviews into either brief snippets or a longer p­ aragraph that
highlights what customers frequently say about a product.4 Amazon is also
beginning to use GenAI to summarize customer reviews into one digestible
paragraph.5

The key takeaway here is that GenAI can enhance the shopping journey by
adding new, more intuitive, and more personalized features that increase cus-
tomer satisfaction and help find exactly the right product for them.

The rise of virtual try-­on features

The online shopping experience can often fall short because you can’t try
something before you buy it. Determining what an item of clothing will look
like on you (instead of the model), or visualizing what an item of furniture
would look like in your house can be a total guessing game. And those issues
apply just as much to glasses, beauty items, and other products.

Until now, virtual try-­on features have been fairly rudimentary – for example,
picking from a small list of body shapes to see what a particular dress might
look like on your body. But thanks to augmented reality (AR) technology,
virtual try-­ons are getting better. Typically, this is done using a phone camera,
but it can also be done using “smart mirrors” (more on these coming up). The
AR technology then layers digital elements (say, glasses, or an item of cloth-
ing) over the real-­world image in the camera.

GenAI will bring a new level of realism to such features. Imagine creating a
realistic avatar of yourself, based on a photo and your exact measurements,
and then being able to generate an image of yourself wearing a particular
outfit (even combining different items to see how well they go together), and
seeing yourself wearing those items in a range of different settings, all to help
you fully visualize whether the product’s right for you. GenAI makes all that
possible. Having an item arrive only to discover it’s not right could be a thing
of the past. For retailers, this means lower return rates and higher customer
satisfaction.

123
 GENERATIVE AI IN PRACTICE

Some of the brands experimenting with virtual try-­on features include Sephora
(digitally superimposing makeup shades onto the user’s face), ­Ray-­Ban (for
eyewear), and Baume & Mercier (for luxury watches).6

In fashion, start-­up Body Labs (owned by Amazon) has created mixed reality
technology that enables shoppers to “see” themselves wearing virtual clothing
in a virtual setting, such as a formal event or the beach.7 Google has also intro-
duced virtual try-­on technology that uses GenAI to show clothes “on a wide
selection of real models.”8 Google says the GenAI model can take just one
picture of an item of clothing and accurately show how it would cling, drape,
etc. on a diverse range of real models from sizes XXS to 4XL, and representing
different skin tones and ethnicities. The feature is available for brands such as
Anthropologie and H&M.

Even furniture and home décor can benefit from virtual try-­before-­you-­buy
features powered by GenAI. Wayfair, for example, has introduced a new
GenAI feature called “Decorify,” which creates photorealistic images that
help customers see their homes in new styles (choosing from themes such
­Bohemian and Farmhouse).9 Users can then shop the Wayfair catalog directly
from their reimagined room.

By the way, virtual try-­on features can also apply to the physical, in-­store
shopping experience. Which brings us to. . .

Augmenting Physical Stores

The in-­store experience could also be enhanced with GenAI technology,
­particularly when it comes to creating interactive store displays and smart
mirrors for virtual try-­on experiences in stores.

GenAI can be used to create responsive store displays that change based on
real-­time data. Say, for example, a cold front is moving into an area, a cloth-
ing store’s digital window display might adapt to showcase winter gear and
generate a live snowfall backdrop to attract people in. Combined with per-
sonalized customer data, in the future, store displays could even adapt to

124
RETAIL REIMAGINED: THE RISE OF VIRTUAL TRY-­O NS, AI SHOPPING ASSISTANTS, AND MORE

each customer, showing them products they might be interested in locating

in the store.

As for smart mirrors (or smart fitting rooms), GenAI combined with AR tech-
nology could ensure that a digital representation of clothing on a customer is
accurate, giving a photorealistic visualization of the fit and look. Why would
we want to virtually try on clothes when we’ve traveled to a physical store?
Well, the store in question may not have all styles, colors, and sizes available
in the store to try on. Before they order your particular item, wouldn’t it be
nice to see how it might realistically look on you?

I mentioned earlier that beauty brand Sephora is using virtual try-­on tech-
nology; the company has also installed kiosks in stores that offer the same
try-­on experience for customers visiting physical stores. Handbag and acces-
sory brand Coach has also experimented with smart mirror technology. To
celebrate the launch of its Tabby Bag campaign, the brand installed a smart
mirror in its Soho New York store.10 Customers using the mirror could see
themselves with different digital variations of the bag (and other digital
effects, like butterfly wings). They could also download their image to share
online. So, as an added bonus, Coach benefited from free social marketing
thanks to users sharing their images on social media.

Even when trying on a physical, real item of clothing, AI-­powered mirrors

could still add value for customers. They could, for example, provide real-­
time feedback, suggesting sizes, colors, or alternative styles based on the
clothes being tried on. The mirror might even generate virtual previews of
how different accessories or shoes would complement the chosen outfit.

Bottom line, while GenAI is set to enhance the digital shopping experience, it
could also give us a more engaging real-­world shopping experience.

Personalizing the Customer Journey

Once again, personalization crops up as a major use case for GenAI. This
applies to all aspects of retail, but let’s explore a few specific ways retailers can
provide a more personalized shopping experience.

125
 GENERATIVE AI IN PRACTICE

Personalized, unique product designs

No longer confined to a one-­size-­fits-­all approach, retailers can employ GenAI
to craft custom product designs based on individual consumer behaviors and
preferences. For instance, a fashion brand could generate unique clothing
patterns or accessories that align closely with evolving trends and individ-
ual tastes. Fashion tech specialists Space Runners have created a GenAI tool
that allows people to design their own unique clothing, just using simple text
prompts.11 Called “Ablo,” the AI design tool allows individuals to effectively
become their own fashion brand – and for brands, Ablo allows amazing co-­
creation opportunities with their customers.

Combine this trend with the growing trend of 3D printing, and we have more
opportunities than ever for individuals and brands to create one-­offs that can
be printed without having to worry about economies of scale.

Personalized shopping journeys

Online platforms can use GenAI to redesign the shopping journey dynami-
cally, amending what the user sees on-­the-­fly. For example, as a user browses,
the AI might rearrange product placements based on the customer’s interests,
showcase user-­generated content to encourage longer engagement, or even
change the website’s color scheme to resonate with the shopper’s preference.

According to research by McKinsey, 7 out of 10 consumers expect brands to

personalize the content they see, such as ads and recommendations, and 76%
get frustrated when they don’t feel their experience is personalized.12

Amazon and Netflix’s recommendation engines are well-­known examples of

personalizing the customer journey. But with GenAI, we can expect this per-
sonalization to become more intelligent. Instead of simply grouping custom-
ers into personas based on interests or past habits (e.g., “people who bought
X”), AI-­based personalization can deliver information and recommendations
based on the individual, not a generic persona.

We’re talking about true, one-­

on-­
one personalization, in other words.
Remember Newegg from earlier in the chapter? As a way to personalize the

126
RETAIL REIMAGINED: THE RISE OF VIRTUAL TRY-­O NS, AI SHOPPING ASSISTANTS, AND MORE

customer journey, the online retailer has integrated ChatGPT into its “PC
Builder” online shopping tool, which allows customers to build a customer
desktop PC system. The new, improved functionality allows customers to
type their desired attributes, and then ChatGPT will make recommenda-
tions.13 We’ll see more and more retailers introduce features like this, where
the shopper can describe what they’re looking for, and a chat AI will be able to
make relevant recommendations. And that brings us to the notion of personal
shopping assistants. . .

Virtual AI shopping assistants

Product recommendations with personality – that’s what GenAI brings to the
customer journey. Expedia’s use of ChatGPT to help travelers plan the per-
fect journey is a great example of using GenAI to create what’s effectively a
virtual, personal shopping assistant. We’re talking, then, about conversational
bots that help customers through the shopping journey, giving thoughtful
responses to their queries and recommending products.

French retailer Carrefour has done this, integrating an advice robot called
“Hopla” (based on ChatGPT) into its website.14 Customers can ask the bot
natural language questions that help them complete their shopping – for
example, which products are best for their budget, food constraints, or menu
ideas, and how to reuse ingredients to save wasting food.

eBay has introduced ShopBot, an AI personal shopping assistant available (at

the time of writing) in Facebook Messenger. ShopBot is designed to help cus-
tomers navigate through eBay’s one billion product listings and find the best
deals for them – by typing or saying what they’re looking for, or snapping
a picture. The bot then asks questions to better understand the customer’s
needs, and it makes personalized recommendations. eBay says its goal is to
“make shopping with eBay as easy as talking to a friend, whether you are
looking for something specific or just browsing for inspiration.”15

Meanwhile, Walmart’s forthcoming GenAI shopping assistant (forthcoming

at the time of writing) will be able to help shoppers plan the perfect event.
Say you want to throw a unicorn-­themed party for a six-­year-­old, the tool

127
 GENERATIVE AI IN PRACTICE

will recommend relevant products, such as unicorn-­themed balloons and

­banners – without you having to do multiple individual searches.16

Second-­hand marketplace Mercari has unveiled its own AI shopping assistant

called “Merchat AI,” powered by (you guessed it) ChatGPT.17 Instead of hav-
ing to search and scroll through multiple options, shoppers can simply tell the
AI what they’re looking for and it’ll point them in the right direction.

I believe AI shopping assistants is a major use case for GenAI in retail, and
I expect traditional “search and scroll” experiences will be increasingly
replaced with conversational search and recommendations. As someone
who’s frequently frustrated by the online shopping experience, that sounds
pretty good to me.

What’s more, AI shopping assistants won’t just help with shopping – they’ll be able
to provide more contexts and, in turn, deliver greater value for customers. Take,
for example, the “Ask Instacart” AI search tool unveiled by Instacart. It doesn’t just
help customers with shopping questions and personalized recommendations – it
can also provide extra info about food preparation, dietary considerations, and
more.18 So you can ask it which side dishes go well with your protein of choice, or
what kinds of sauce would work well, or ideas for “free-­from” foods.

Tailored promotions, discounts, and loyalty programs

We talked about this already in terms of hyper-­personalized advertising and
marketing (see Chapter 7), but let me briefly reiterate that GenAI can create
custom promotions and offers for each shopper – by analyzing their individ-
ual buying habits and preferences. For example, if a customer often purchases
sustainable products, the AI might generate special discounts or bundle offers
related to eco-­friendly items during their next visit. This can be done auto-
matically, and at scale.

GenAI can also be used to develop highly personalized loyalty programs that
evolve based on each member’s interactions with the brand. So, instead of
a generic point system, members might be offered personalized challenges,
rewards tailored to their preferences, or unique experiences curated just for them.

128
RETAIL REIMAGINED: THE RISE OF VIRTUAL TRY-­O NS, AI SHOPPING ASSISTANTS, AND MORE

Other Potential Uses in Retail

As well as augmenting the online and real-­world shopping experience, and
enabling a personalized customer journey, GenAI could support retailers in
other innovative ways. It can be used to provide marketing and sales sup-
port through automated and personalized marketing messages (again, see
­Chapter 7). It can upsell customers by telling them about similar or comple-
mentary products. It can be used to manage inventory levels in a more intel-
ligent way, by analyzing sales data and forecasting demand – thereby helping
retailers ensure they stock the right products in the right quantities. Or it can
help customers handle returns and exchanges, answering questions about the
returns process and guiding customers through the process step-­by-­step.

Let’s explore a couple of use cases in a little more detail.

Generating product descriptions

One area in which GenAI excels is in writing descriptions, including bespoke
descriptions that tell a story about a product. For example, a coffee brand might
create a dynamic story for each coffee blend, detailing its journey from a specific
farm, its unique roasting process, and flavor notes. For companies that empha-
size storytelling as part of their brand identity, GenAI can add huge value.

The same goes for generic product descriptions. A retailer or online seller might
not want to craft a compelling story for each individual product, but simply want
to streamline the process of creating informative, search-­friendly descriptions.

eBay is helping sellers do just that with a new GenAI “magical listing” tool
that writes product listings from photos.19 So, a seller can simply take or
upload a photo in the app and then let the AI complete the product details for
them. Shopify has a similar tool that it calls Shopify Magic – which automates
the content creation process for descriptions, email subject lines, and headers
for an online store.20 Likewise, Amazon has debuted a GenAI tool that helps
sellers create product descriptions.21

Diesel has also turned to AI to enhance its product data, using automated prod-
uct tagging to enhance product discovery online. By switching from manually

129
 GENERATIVE AI IN PRACTICE

tagging products according to various attributes and categories to automated

tagging, the company was able to save 30 hours of time per week per person. 22

Automating supplier negotiations

As well as creating a personal AI shopping assistant, we know that Walmart has
been experimenting with GenAI across internal operations (see ­Chapter 3).
One particularly interesting use case is in supplier negotiations. In an experi-
ment, the retailer used a chatbot to close deals with 89 suppliers of items like
shopping carts and other store equipment. Of those suppliers, the chatbot
closed deals with 64%, gaining an average of 1.5% in cost savings, and an extra
35 days in extended payment terms. An impressive 83% of suppliers actually
liked the chatbot negotiation.23

Deploying GenAI across the organization

Spanish fashion retailer, Mango, has introduced its own conversational GenAI
platform called “Lisa,” which is designed to help employees across the busi-
ness make improvements – and this can span anything from enhancing the
after-­sales service to developing new clothing collections. The tool has been
described as a “co-­pilot for our employees.”24 Mango has also previously devel-
oped a GenAI image platform called “Inspire,” aimed at helping the design
team find inspiration and come up with new concepts for prints, fabrics, and
garments. But we’ll talk more about GenAI in the design process in Chapter 14.

Key Takeaways
Make no mistake, retail will be utterly transformed by GenAI tools. To recap
some of the main use cases from retail:

• GenAI enables more immersive digital experiences – for example,

through a dynamic, responsive customer journey in metaverse shops.
But it will also enable a better shopping experience on regular websites
and apps, with realistic virtual-­on features and handy summaries of
customer reviews.

130
RETAIL REIMAGINED: THE RISE OF VIRTUAL TRY-­O NS, AI SHOPPING ASSISTANTS, AND MORE

• GenAI will augment the physical in-­store experience, too, thanks to

dynamic displays and smart mirrors.
• Once again, personalization is a huge theme. GenAI enables hyper-­
personalization for retailers in a number of ways, including one-­off
products, personalized shopping journeys, AI shopping assistants, and
tailored promotions.
• Other uses in retail include automatically generating product descrip-
tions, automating aspects of the supply chain, and even automating
supplier negotiations.

Now let’s go back to school and see how GenAI will transform how people
learn, what people learn, and how educators teach.

Notes
1. What Will Fashion Retail Look Like in the Future?; Forbes; https://www.forbes
.com/sites/bernardmarr/2023/07/13/what-­w ill-­f ashion-­retail-­look-­like-­in-­
the-­future/
2. What Will Fashion Retail Look Like in the Future?; Forbes; https://www.forbes
.com/sites/bernardmarr/2023/07/13/what-­w ill-­f ashion-­retail-­look-­like-­in-­
the-­future/
3. The Amazing Ways Expedia Is Using ChatGPT to Simplify Travel Arrange-
ments; Bernard Marr; https://bernardmarr.com/the-­amazing-­ways-­expedia-­is-­
using-­chatgpt-­to-­simplify-­travel-­arrangements/
4. Newegg adds ChatGPT-­powered feature to save you from review analysis paral-
ysis; ZDNet; https://www.zdnet.com/article/neweggs-­ai-­review-­summary-­tool-­
could-­make-­shopping-­for-­electronics-­a-­little-­easier/
5. Amazon now using generative AI to summarize customer reviews; ZDNet; https://
www.zdnet.com/article/amazon-­n ow-­u sing-­g enerative-­ai-­t o-­s ummarize-­
customer-­reviews/
6. 7 Brands Using Virtual Try-­on Tech to Boost Sales; ThreeKit; https://www.threekit
.com/blog/7-­brands-­using-­virtual-­try-­on-­boost-­sales
7. Virtual Try-­on for Clothing: The Future of Fashion?; 3D Look; https://3dlook.ai/
content-­hub/virtual-­clothing-­try-­on/
8. Virtually try on clothes with a new AI shopping feature; Google; https://blog
.google/products/shopping/ai-­virtual-­try-­on-­google-­shopping/
9. Shop Virtually with Wayfair’s Generative AI Decorify; RetailWire; https://
retailwire.com/discussion/shop-­virtually-­with-­wayfairs-­generative-­ai-­decorify/

131
 GENERATIVE AI IN PRACTICE

10. Why the New Generation of AR Smart Mirrors Are Catnip for Fashion and
Beauty Retail; Forbes; https://www.forbes.com/sites/stephaniehirschmiller/2023
/05/19/how-­ar-­mirrors-­offer-­visual-­marketing-­meets-­user-­generated-­content-­
for-­fashion-­and-­beauty-­retail/
11. Collaborative Fashion Innovation: Unlocking Creativity with AI; Medium;
https://medium.com/@spacerunners/collaborative-­f ashion-­i nnovation-­
unlocking-­creativity-­with-­ai-­91b424d9e5ba
12. The value of getting personalization right – or wrong – is multiplying; McKinsey;
https://www.mckinsey.com/capabilities/growth-­m arketing-­and-­s ales/our-­
insights/the-­value-­of-­getting-­personalization-­right-­or-­wrong-­is-­multiplying
13. Newegg Uses ChatGPT to Improve Online Shopping Experience; Business Wire;
https://www.businesswire.com/news/home/20230327005151/en/Newegg-­Uses-­
ChatGPT-­to-­Improve-­Online-­Shopping-­Experience
14. Carrefour integrates OpenAI technologies and launches a generative AI-­
powered shopping experience; Carrefour; https://www.carrefour.com/en/news/
2023/carrefour-­integrates-­openai-­technologies-­and-­launches-­generative-­ai-­
powered-­shopping
15. Say ‘Hello’ to eBay ShopBot Beta; eBay; https://www.ebayinc.com/stories/news/
say-­hello-­to-­ebay-­shopbot-­beta/
16. Walmart experiments with generative AI tools that can help you plan a party or deco-
rate; TechCrunch; https://techcrunch.com/2023/10/04/walmart-­experiments-­with-­
new-­generative-­ai-­tools-­that-­can-­help-­you-­plan-­a-­party-­or-­decorate-­a-­space/
17. ChatGPT can help you shop now via Mercari’s AI-­powered assistant; ZDNet;
https://www.zdnet.com/article/chatgpt-­can-­now-­help-­you-­shop-­via-­mercaris-­
new-­ai-­powered-­shopping-­assistant/
18. Instacart launches new in-­app AI search tool powered by ChatGPT; TechCrunch;
https://techcrunch.com/2023/05/31/instacart-­in-­app-­ai-­search-­tool-­powered-­
by-­chatgpt/
19. eBay’s new AI tool generates product listings from photos; Retail Dive; https://
www.retaildive.com/news/ebay-­a i-­m agical-­l isting-­product-­d escriptions-­
listings/693185/
20. Shopify Magic; Shopify; https://help.shopify.com/en/manual/shopify-­admin/
productivity-­tools/shopify-­magic
21. Amazon debuts generative AI tool that helps sellers write product descriptions;
TechCrunch; https://techcrunch.com/2023/09/13/amazon-­debuts-­generative-­ai-­
tools-­that-­helps-­sellers-­write-­product-­descriptions/
22. Product tagging for Diesel; Vue AI; https://vue.ai/resources/case-­studies/
product-­tagging-­for-­diesel/
23. The Supply Side: Generative AI or ChatGPT next holy grail for retailers, suppli-
ers; Talk Business & Politics; https://talkbusiness.net/2023/06/the-­supply-­side-­
generative-­ai-­or-­chatgpt-­next-­holy-­grail-­for-­retailers-­suppliers/
24. Mango launches conversational generative AI platform; Fashion United; https://
fashionunited.uk/news/business/mango-­launches-­conversational-­generative-­
ai-­platform/2023100571951
132
 10
PERSONALIZED
LEARNING: THE FUTURE
OF EDUCATION

AI has been making inroads into education for the last few years, and as with
many sectors, GenAI will massively accelerate this trend. Indeed, it could be a
game-­changer, having a profound effect on how teachers teach and how stu-
dents learn – not to mention what students learn. But GenAI comes with risks
and challenges for educators. How can we make sure students don’t cheat on
assessments and homework? How should educators prepare for this new era
of on-­demand, personalized AI tools?

It’s still early days for GenAI in education, but let’s paint a picture of where
the sector is headed.

Rethinking How Learners Learn, and

How Teachers Teach
In any education setting, from schools and colleges to self-­directed lifelong
learning, GenAI has the capacity to provide super-­personalized learning,
make teachers’ jobs easier, and help students understand topics more easily.
As such, we may see vast changes in how students learn and how teachers
teach. We’ll talk more about what this means for education policy later in the
chapter, but for now let’s explore the practical uses of GenAI in education.

133
 GENERATIVE AI IN PRACTICE

How GenAI can add value for educators and students

Arguably the biggest advantage of GenAI is the ability to provide a more per-
sonalized learning experience. One of the many challenges teachers face is
identifying where each student needs additional support, and then providing
that support in a busy classroom setting (or, indeed, in a remote learning envi-
ronment). GenAI can assist with this by creating assessments and questions
in real time – giving educators more feedback to help them identify how stu-
dents are progressing and pinpoint which students might be struggling with
a concept. And, of course, GenAI can help to provide the extra support when
needed. For instance, when a student struggles with a math problem, a GenAI
tutor can explain the concept step-­by-­step, adapting its explanations to the stu-
dent’s level of understanding. GenAI can also be used to generate personalized
­follow-­up questions to help cement understanding. Plus, GenAI can create indi-
vidual homework plans for each student. We’re talking about adaptive learning,
­basically – learning that is tailored to each student’s needs, and to their pace.

Another significant application is in content creation. GenAI can generate

educational materials such as quizzes, interactive games, flashcards, lesson
plans, and even entire textbooks or video courses, making it easier for educa-
tors to create engaging and relevant learning resources for use in class. And
again, these materials can be easily personalized if desired.

Moreover, language models like GPT-­4 have enabled AI-­powered chatbots that
can answer students’ questions in real time, enhancing accessibility and support
for students (especially for remote and online learners). As the following exam-
ples will show, GenAI can be used to create dedicated virtual tutors that are
available 24/7, allowing students to learn entirely at their own pace. And even
outside of formal classroom settings or online learning platforms, GenAI is
already making its way into education. With developments like ChatGPT being
integrated into Snapchat, children are already beginning to access GenAI, ask it
questions, and use it as part of their learning, even in informal ways.

Overall, then, GenAI will revolutionize education by enhancing personali-

zation, accessibility, and efficiency in the learning process. No wonder the
UK government has said GenAI has the potential to “reduce workload across
the education sector” and “free up teachers’ time, allowing them to focus on

134
 PERSONALIZED LEARNING: THE FUTURE OF EDUCATION

delivering excellent teaching.”1 The UK Department for Education says it is

working with the education sector to “identify opportunities to improve edu-
cation and reduce workload using generative AI.” More on this coming up.

Helping teachers teach

Now let’s explore some of the many GenAI tools designed specifically for
education. MagicSchool is just one example of a GenAI tool designed for
teachers. Based on OpenAI technology, the tool can be used to create word
problems, worksheets, tests, and lesson plans. It can also adjust the reading
level of materials based on individual students’ needs, advise teachers on how
to deal with behavioral problems in class, and more. Promising to save teach-
ers time, reduce stress, and avoid burnout, tools like this will become increas-
ingly popular among teachers. Indeed, MagicSchool amassed an impressive
150 000 users in its first four months.2

Kortext is another interesting example. The education platform is designed to

help universities and colleges provide AI-­enhanced teaching and personalized
learning experiences. Kortext tools can instantly summarize content, generate
study notes, and create interactive Q&As to cement learning. It can also deliver
valuable data to course leaders on student engagement, study patterns, and con-
tent usage. And, importantly, unlike many other GenAI tools, Kortext uses only
trusted content (textbooks, etc.) that has been prescribed as part of the institu-
tion’s own curriculum. Among the thousands of universities using Kortext is
Middlesex University, University of Manchester, and University of Oxford.3

GenAI can also be used internally to help staff and students with daily admin
tasks. In one example, Purdue University says it has used GenAI to give tech-
nical advice – for example, explaining how to set up a networked printer –
and has plans to use the technology for other advice, such as how to find
buildings on campus.4

Helping learners learn

Meanwhile, students can benefit from GenAI-­powered learning tools such
as Chegg, which is designed to facilitate 24/7 study and help learners access

135
 GENERATIVE AI IN PRACTICE

instant, step-­by-­step explanations on any subject or question. According to

Chegg, 94% of its users say they get better grades when they use Chegg to
understand their coursework.5

There’s also Khan Academy, a nonprofit organization on a mission to pro-

vide free education for anyone, anywhere. Khan Academy has partnered with
OpenAI and is using the GPT-­4 language model to create Khanmigo, a virtual
tutor for students that can, among other things, ask each student individual-
ized questions to prompt deeper learning.6 (Khan Academy is also exploring
using GPT-­4 as a classroom assistant for teachers – e.g., creating instructional
materials and classroom prompts.)

Udacity is another online course provider using GPT-­4 to create a virtual,

on-­demand, AI tutor – a tutor that can provide detailed explanations tailored
to each individual learner, summarize concepts, and even translate learning
materials into other languages. Udacity says the chatbot tutor is a “real-­time
complement to our human mentors,” not a replacement for human mentors.7
And importantly, the AI tutor can handle thousands of interactions at once.

There’s also Quizlet’s GenAI study aid, which provides personalized AI tutor-
ing, automatic flashcards, quick summaries of concepts, and questions that
help students understand topics (instead of just telling them the answer). Over
a million students have used Quizlet’s Q-­Chat chatbot, and Quizlet research
shows that more than two-­thirds of students believe the AI helps them better
understand materials and learn faster.8

And for lifelong learners whose school days are long behind them, GenAI
can make learning any topic easier. Take learning a second language as an
example. Duolingo is the poster child for using AI in language learning, and
has collaborated with OpenAI to integrate GPT-­4 into its offering.9 The result
is a much more personalized language-­learning experience. In particular, the
partnership has introduced two new features to Duolingo: the first feature,
called “Explain My Answer,” provides users with detailed explanations of
their responses, mimicking the feedback you might get from a human lan-
guage tutor; and the second “Role Playing” feature allows users to engage with
AI personas for immersive language practice in diverse scenarios. For exam-
ple, say you’re planning a trip to Paris and have specific dietary requirements;

136
 PERSONALIZED LEARNING: THE FUTURE OF EDUCATION

you can tailor your role-­playing lessons to have conversations with an AI

barista. The AI personas are designed to be fully fleshed-­out characters with
their own personality and backstory that learners can uncover through con-
versations. With these new additions, Duolingo is able to cater to each of its
500-­million-­plus students as individuals, which is an impressive feat.

Transforming What We Teach

As well as transforming how subjects are taught, the rise of GenAI may
change what subjects are taught. We talked about this a little in Chapter 5 –
about how the important skills for success will be different in a world where
machines can do more and more jobs. This is the subject of my book, Future
Skills: The 20 skills and competencies everyone needs to succeed in a digital
world, and I recommend you seek it out if you’d like a more in-­depth look at
skills. But from a formal learning perspective, we may see institutions incor-
porate subjects and skills that are complementary to machines in our increas-
ingly AI-­driven world. Skills like creative thinking, empathy, and teamwork.

We will also need to teach young people how to critically assess the informa-
tion they access, especially given the huge potential for AI-­generated misin-
formation. For example, students will need to learn to search for clues in text
to determine the authenticity and accuracy of the text. They will need guid-
ance on how to gather evidence – and what constitutes good evidence – in a
world where you can just ask a chatbot for information and immediately be
given an answer.

And of course, students will need to learn how to be digitally literate. I’m not
talking about teaching every young person how to code; rather, I’m talking
about teaching every young person about AI and its impact on our world.
That includes understanding the potential for misinformation, understand-
ing the potential for data bias, understanding how AIs are trained (and how
that impacts the information they serve up), and understanding how AI
dictates so much of the content we see online (leading to information bub-
bles). I would love to see young people learning how to thrive in a digital
world, and how humanity can get the best out of AI without losing what
makes us human.

137
 GENERATIVE AI IN PRACTICE

Given its transformational potential, we may see GenAI integrated into core
curricula around the world. I certainly hope so. As more and more jobs will be
augmented by GenAI, it makes sense that students should learn how to har-
ness GenAI to solve problems, enhance their productivity, and boost innova-
tion. The California Department of Education is one authority “encouraging”
districts to explore the potential benefits of AI as part of the curriculum.10 But
we need more formal frameworks than that – not just in terms of including
AI in the curriculum, but also how to harness GenAI properly in education
settings. Which brings us to. . .

The Need for Formal Policy and Guidance

in Schools
It should be clear by now that I’m broadly in favor of schools, colleges, and
individual students embracing GenAI. But GenAI isn’t without its challenges.
For one thing, we need teachers (and students) to be aware of potential bias
and misinformation in educational content generated by AI. We need institu-
tions to ensure GenAI is inclusive. We need to use GenAI as a complement to
human teachers, not a replacement. And we may even need to rethink how
students are assessed – cheating being a big concern among some educators.
Say, for example, a student is wearing a smart watch in an exam, what’s to stop
them having ChatGPT dictate answers to them via unseen wireless earbuds?
At the very least, there’s the danger of students relying on ChatGPT to write
their homework for them. So how can educators prepare for this – and pre-
pare their students to use GenAI responsibly?

Of course, institutions faced similar conversations when calculators and, later,

computers were introduced, but the question remains: How can we encourage
students to use these GenAI tools effectively but not rely on them?

Despite teachers and students already beginning to experiment with GenAI

tools, the education sector remains largely unprepared for the rise of GenAI.
A UNESCO study of over 450 schools and universities around the world
found that less than 10% had institutional policies or formal guidance in
place outlining the use of GenAI applications – largely due to lack of govern-
ment frameworks.11 This has led UNESCO to call for governments to provide

138
 PERSONALIZED LEARNING: THE FUTURE OF EDUCATION

formal regulation and teacher training on the use of GenAI in education set-
tings. UNESCO says the lack of scrutiny, checks or regulations on the use of
GenAI is worrying, especially given the scrutiny applied to other aspects of
education, such as which books are taught in the classroom.

Some authorities are beginning to grapple with AI policy. New York City
­Public Schools, for instance, has collaborated with AI experts to launch
an AI Policy Lab to guide the school district’s approach to AI. The Policy
Lab will focus on “human-­centered AI implementation, equity, safety, eth-
ics, effectiveness, and transparency.”12 That’s an encouraging and pragmatic
turnaround, considering the district initially banned the use of ChatGPT on
school networks.

The use of GenAI in college applications is another hot topic. Common App,
the college application tool used by more than 1000 institutions, has included
a restriction on the use of “substantive content or output of an artificial intel-
ligence platform” in college applications.13 (Although, my question is, how on
earth do they police this? How do they intend to identify when a “substantive”
amount of content has been written by GenAI?)

Here in the UK, I’ve already mentioned that the UK Department for Education
(DfE) is keen to explore the use of GenAI in schools. At the time of writing, the
DfE had launched a “call for evidence” to gather views from education profes-
sionals on the ethical considerations, risks and possibilities of AI, and invited
teachers to participate in a “hackathon” event designed to explore AI’s potential
across different scenarios. Results were due to be published at the end of 2023.14

Also in the UK, the Russell Group, comprising 24 universities, has drawn
up “guiding principles” for the use of GenAI in its universities – designed
to help universities capitalize on the technology while maintaining academic
rigor. The new policies “make it clear to students and staff where the use of
generative AI is inappropriate, and are intended to support them in making
informed decisions and to empower them to use these tools appropriately and
acknowledge their use where necessary.”15

We’ll no doubt see more education institutions around the world begin to
account for GenAI in their policies and guidance, but this cannot happen

139
 GENERATIVE AI IN PRACTICE

soon enough. At present, the use of GenAI in education is reminiscent of the

old Wild West. For now, anything goes. . .

Key Takeaways
In this chapter, we’ve learned:

• GenAI is revolutionizing education by enhancing personalization,

accessibility, and efficiency in the learning process. Adaptive, on-­
demand tutoring is a particularly hot area. The use of GenAI in educa-
tion promises not only to help students learn, but also to help teachers
teach – freeing up teachers’ time to focus on students rather than, say,
writing lesson plans and devising tests.
• The rise of GenAI may also influence what is taught in institutions.
From school to university level, students will need to be equipped with
the skills and knowledge to thrive in an AI-­driven world.
• At present, few institutions have formal policies on how GenAI should
be used by teachers and students, prompting concerns about cheating
and the potential for misinformation and bias in educational content.
This is something education providers need to think about, and fast!

I like talking about education because it’s a subject that’s close to many of our
hearts – we’ve all been learners at one time or another and many of us have
children in education. Now let’s turn to another sector with universal interest:
healthcare. How will GenAI transform the healthcare sector and, hopefully,
improve healthcare outcomes for us all? That’s the subject of our next chapter.

Notes
1. Generative artificial intelligence (AI) in education; Department for Education;
https://www.gov.uk/government/publications/generative-­artificial-­intelligence-
­in-­education/generative-­artificial-­intelligence-­ai-­in-­education#:~:text=Generative
%20AI%20refers%20to%20technology,large%20language%20models%20(%20
LLMs%20)

140
 PERSONALIZED LEARNING: THE FUTURE OF EDUCATION

2. Teachers Are Going All In on Generative AI; Wired; https://www.wired.com/

story/teachers-­are-­going-­all-­in-­on-­generative-­ai/
3. Kortext; Kortext; https://www.kortext.com/
4. AI Buzz Dominates Annual Ed-­Tech Conference; Inside Higher Ed; https://www
.insidehighered.com/news/tech-­innovation/artificial-­intelligence/2023/10/12/
ai-­buzz-­dominates-­annual-­ed-­tech-­conference
5. Chegg; Chegg; https://www.chegg.com/
6. Khan Academy; OpenAI; https://openai.com/customer-­stories/khan-­academy
7. Announcing our new AI chatbot; Udacity; https://www.udacity.com/blog/
2023/03/announcing-­our-­new-­ai-­chatbot-­on-­d emand-­l earning-­g uidance-­
made-­easy.html
8. Quizlet is using AI to make studying easier – with personalized tutoring, auto-
matic flashcards, and even songs; Fortune; https://fortune.com/education/
articles/quizlet-­ai-­powered-­tools-­q-­chat-­magic-­notes-­quick-­summary-­gpt/
9. The Amazing Ways Duolingo Is Using AI and GPT-­4; Forbes; https://www
.forbes.com/sites/bernardmarr/2023/04/28/the-­amazing-­ways-­duolingo-­is-­
using-­ai-­and-­gpt-­4/
10. ‘We Can’t Pretend It Away’: Schools Grapple with Guidelines for Using Generative
AI on College Applications; KQED; https://www.kqed.org/news/11964259/we-­
cant-­pretend-­it-­away-­schools-­grapple-­with-­guidelines-­for-­using-­generative-­ai-­
on-­college-­applications#:~:text=The%20California%20Department%20of%20
Education,broader%20lessons%20in%20media%20literacy.
11. UNESCO: Governments must quickly regulate generative AI in schools; UNESCO;
https://www.unesco.org/en/articles/unesco-­governments-­must-­quickly-­regulate-­
generative-­ai-­schools
12. NYC Schools Working with AI Experts to Launch AI Policy Lab; Government
Technology; https://www.govtech.com/education/k-­12/nyc-­schools-­working-­
with-­experts-­to-­launch-­ai-­policy-­lab
13. Application Fraud; Common App; https://www.commonapp.org/files/Common-­
App-­Fraud-­Policy.pdf
14. Teachers to join DfE’s AI workload reduction trial; TES Magazine; https://
www.tes.com/magazine/news/general/artificial-­intelligence-­reduce-­teacher-­
workload-­government-­trial
15. UK universities draw up guiding principles on generative AI; The Guardian;
https://www.theguardian.com/technology/2023/jul/04/uk-­universities-­draw-­
up-­guiding-­principles-­on-­generative-­ai

141
 11
HEALTHCARE
TRANSFORMATION:
FROM PERSONALIZED
ADVICE TO OPERATIONAL
IMPROVEMENTS

GenAI is ushering in a transformative era in healthcare, with far-­reaching

implications for patient care, diagnostics, and more. There’s a lot happening
in this field, so this chapter touches on key developments only, with a particu-
lar focus on:

• Personalized advice for patients

• Detection of diseases
• Personalized treatment
• And operational improvements

We’ll return to healthcare in Chapter 19, where we’ll look at the use of GenAI
in drug discovery and development. But for now, let’s get into the amazing
ways GenAI can improve healthcare and deliver better outcomes for patients
(not to mention a better working experience for healthcare professionals).

143
 GENERATIVE AI IN PRACTICE

Giving Patients Personalized Health Advice

This goes beyond basic chatbots that can help patients book appointments,
refill prescriptions, or answer routine enquiries (although such chatbots do
have a valuable place in healthcare, as we’ll see later in the chapter). Sophis-
ticated large language models like GPT-­4, combined with human doctor
expertise, has led to a new wave of virtual health assistants – developed
to give actual medical advice that’s personalized to each patient’s specific
health concerns. With millions of people around the world unable to access
medical care (either because of their geographical location, for economic
reasons, or simply because their local services are too stretched), we can
expect to see AI increasingly pick up the slack. These AI systems not only
improve patient engagement but also relieve the burden on healthcare pro-
fessionals by handling routine healthcare concerns and directing patients
accordingly.

As an example, Ada is a doctor-­developed, AI-­driven app designed to assess

symptoms and offer patients medical guidance, in multiple languages (includ-
ing English, German, French, Spanish, Portuguese, and Swahili). So far, the
app has amassed 13 million users and has completed more than 30 million
symptom assessments.1 It works by asking you questions about your symp-
toms (you can also create separate symptom profiles for loved ones), then
pointing you to possible conditions and medical guidance. The app also
tracks your symptoms as they progress.

Image Analysis and the Early Detection

of Diseases
AI has been making a mark in this area for a little while, but GenAI will sig-
nificantly enhance the analysis of medical images. As such, we’ll increasingly
see GenAI tools being used to help radiologists identify and diagnose diseases
from X-­rays, MRIs, and CT scans with greater accuracy and speed.

But can an AI really be as effective as a human doctor when it comes to detect-

ing and diagnosing issues from images? Absolutely. Numerous studies have
shown that AI tools deliver comparable accuracy to human doctors – if not

144
HEALTHCARE TRANSFORMATION: FROM PERSONALIZED ADVICE TO OPERATIONAL IMPROVEMENTS

higher accuracy. (After all, machines don’t get tired at the end of a long shift.)
But let’s explore one particular example. . .

Detecting issues in chest radiographs

One study explored the use of AI to interpret chest radiographs and gener-
ate radiograph reports in the emergency department. Since many emergency
departments don’t have 24/7 access to dedicated radiology services, images are
often interpreted by a remote radiologist (known as “teleradiology”) or even
by ER doctors. The study found that the AI tool generated rapid radiograph
interpretations and reports with comparable levels of quality and accuracy
as radiologist reports – and to a higher quality than teleradiology reports.2
In one of the cases, the AI performed even better than a human radiologist,
detecting an issue that the radiologist failed to report. And in another case,
the human radiologist reported that opacities in a patient’s lungs appeared
consistent, while the AI detected that the opacities had worsened – a finding
that could be significant in that patient’s case.

The authors of the study suggest that the short processing time and high
accuracy of AI diagnostics could help to streamline the processing of patients
in emergency departments. This shows that not only can AI help radiologists
perform their work more quickly and effectively – it can also help clinicians
in other departments interpret medical images and accelerate the processing
of patients.

Helping patients self-­examine

GenAI is rapidly emerging as a viable tool in medical diagnostics. But it can
also help patients identify issues and seek early medical intervention. That’s
the idea behind SkinVision, the AI-­powered app for early detection of skin
cancer. SkinVision is a regulated medical service that teaches people how to
self-­examine their skin and determine when they need to take action. And
while SkinVision is quick to say its service is not a medical diagnosis, the
idea is to help people quickly detect signs of the most common kinds of skin
cancer, so that they can seek a formal diagnosis where needed. Research has
shown that the app can detect skin cancer with 95% accuracy.3

145
 GENERATIVE AI IN PRACTICE

Providing Personalized and Improved Treatment

for Patients
GenAI can help doctors enhance patient treatment – by analyzing vast patient
datasets to recommend personalized treatment plans, optimizing medication
dosages, and predicting potential adverse reactions, all based on the individual.
Plus, it can assist in the creation of tailored rehabilitation exercises and therapy
programs, enabling more precise and effective care for individual patients.

GenAI can also help to enhance preventive medicine. For example, clinics
and hospitals could employ GenAI to create personalized health plans based
on a patient’s unique genetic makeup, health history, and lifestyle. This sort
of AI-­driven approach could not only help people lead healthier lives and, in
turn, prevent illness – it would also help patients feel more understood and
cared for as an individual.

Supplementing patient–doctor conversations

Imagine a scenario where a patient consults with a doctor, and a GenAI system
operates in the background, listening, taking notes, and formulating potential
questions for the doctor to ask – based on the patient’s history and symptoms.
Like a cross between a medical chatbot and a diagnostic tool, but designed to
be used in sessions with patients to inform conversations and help the doctor
uncover symptoms and diagnoses. (Indeed, later in the chapter we’ll see how
GenAI is already being used to listen to and summarize patient consultations.)

This sort of AI-­augmented approach could help doctors get the most out of
patient appointments (which can be as brief as 10 minutes), while tailoring
their care to the individual in front of them. Plus, it could prove particularly
valuable in the diagnosis of rare or difficult-­to-­diagnose conditions, such as
endometriosis, which affects 1 in 10 women but has symptoms which are
often dismissed or misdiagnosed.

Helping busy doctors provide precision care

A good example comes from GenAI experts RhythmX AI, who have created
a precision care platform that helps doctors deliver hyper-­personalized care.

146
HEALTHCARE TRANSFORMATION: FROM PERSONALIZED ADVICE TO OPERATIONAL IMPROVEMENTS

In essence, the system uses GenAI and predictive AI algorithms to provide

patient-­
specific prescriptive actions and recommendations. Doctors can
then drill into the recommendations via the natural language interface – an
AI co-­pilot for doctors, if you will. As Dr Gregg Meyer, part of the clinical
advisory board at RhythmX put it, “Physicians are now considering clinical
experience, social determinants, lifestyle factors, and mental health factors
to deliver hyper-­personalized interventions . . . There is a critical need in the
industry for a platform that can aid physicians to solve these issues.”4

Healthcare Research and the Development

of New Drugs
We’ll talk more about this in Chapter 19, which is all about the use of GenAI
to design the products and solutions of the future. But I just want to flag
here that the implications of GenAI for healthcare research and drug dis-
covery are huge. GenAI can help researchers understand disease markers
more easily, and find optimal combinations of chemicals (and even invent
entirely new combinations) to create new pharmaceutical treatments. As
such, GenAI will revolutionize drug discovery and development by gener-
ating novel molecular structures, swiftly screening compounds, predicting
drug interactions, repurposing existing drugs for new applications, optimiz-
ing clinical trials, and enhancing drug formulations. Not only will this help
clinicians treat diseases better in future, it will also make treatment more
personalized – because drugs can, in theory, be tailored based on individual
patient data.

Delivering Administrative and Operational

Improvements
GenAI can help to reduce the administrative burden in healthcare by auto-
mating tasks such as medical coding, billing, routine enquiries, and notetak-
ing. If we think about the tasks that GenAI is capable of – writing, listening,
and interpreting human speech and text – it’s clear that GenAI can add a
lot of value behind the scenes in clinical settings. Even just a few minutes
saved on each patient encounter with automated notetaking could add up to

147
 GENERATIVE AI IN PRACTICE

considerable operational improvements – freeing up healthcare professionals

to spend more time with patients and less time on admin tasks.

Conversational AI for answering routine calls

and messages
Hyro is a HIPAA-­compliant “conversational AI” designed to help clinics auto-
mate common interactions, overcome staffing shortages, and enhance patient
engagement. Using natural language text and speech, Hyro helps healthcare
teams automate straightforward tasks and requests from patients, including
appointment scheduling and prescription refills – meanwhile, more complex
cases are routed to the appropriate department. Hyro says its AI assistants can
resolve around 85% of calls.5

Using GenAI for notetaking and clinical documentation

Another good example comes from NextGen Healthcare and its Ambient
Assist notetaking tool, which listens to conversations between patients and
clinicians, then provides summary notes. Notes are available for the clinician
to review within just 30 seconds of completing the patient encounter, with the
tool documenting appointments with over 90% accuracy.6

A 2017 study found that primary care doctors spend nearly half their working
day on electronic health record and admin tasks, and just 27% of their time
with patients – with admin burden being cited as a leading cause of clinician
burnout.7 Tools like NextGen help clinicians cut down on admin tasks with-
out compromising clinical records.

In another example, Baptist Health, a South Florida health system comprising

11 hospitals, is also using GenAI to reduce the time spent on clinical docu-
mentation. It is implementing a GenAI-­powered documentation app that
blends medical transcription technology with AI language models – mean-
ing the system can transcribe patient conversations and quickly generate
clinical notes from the conversations. The automation is expected to reduce
documentation time down to around two minutes post-­visit, giving clinicians
more time to attend to patients.8

148
HEALTHCARE TRANSFORMATION: FROM PERSONALIZED ADVICE TO OPERATIONAL IMPROVEMENTS

Helping clinicians search information more easily

GenAI isn’t just about generating content for clinicians; it can also help clini-
cians access and digest information more easily. That’s the idea behind Google
Cloud’s Vertex AI Search platform, which has now been tailored specifi-
cally for healthcare and life-­sciences organizations. Using the tool, clinicians
can search for information across data sources – not just electronic health
records, but potentially the organization’s entire data ecosystem, including
raw unstructured data. Among the companies integrating Google Cloud’s
GenAI search capabilities is care.ai. The AI company says its ‘Smart Care
Facility Platform’ will use Google’s GenAI technology to enable new ways to
analyze data, generate insights, and enhance care delivery.9

Predicting outbreaks to optimize resources

Another way that AI may streamline healthcare provision is in forecasting
disease outbreaks, or predicting how viruses may change. An exciting exam-
ple comes from an AI tool called “EVEscape,” developed by researchers at
Harvard Medical School and the University of Oxford. EVEscape uses gen-
erative models to predict how a virus could evolve to escape the immune
system, thereby predicting new mutations of viruses. (The tool successfully
predicted the most concerning new variants of COVID-­19.10) Not only could
this help global health systems prepare for and manage pandemics, it could
also inform the development of vaccines and therapies for rapidly chang-
ing viruses.

Could Generative AI Help to Solve

the Healthcare Crisis?
Around the world, healthcare systems are coming under enormous strain.
In many regions, there is a critical shortage of doctors and healthcare pro-
fessionals. The World Health Organization (WHO) estimates a projected
shortfall of 10 million healthcare professionals by 2030, primarily affecting
low-­ and lower-­middle-­income countries.11 Even in a rich country like the
US, the Association of American Medical Colleges is predicting a shortage of
up to 139 000 physicians by 2033.12

149
 GENERATIVE AI IN PRACTICE

What’s more, healthcare systems are plagued by rising waiting times. At the time
of writing, the waiting list for planned treatment in England’s National Health
Service hit another record high of 7.75 million patients, with thousands waiting
more than 18 months to start treatment.13 And that’s the situation in a country
with universal healthcare – something that many people aren’t fortunate enough
to enjoy. Shockingly, WHO figures from 2021 indicate that around 4.5 billion
people around the world are not fully covered by essential health services.14

The promise of GenAI

There’s no substitute for receiving amazing care from human doctors and other
healthcare professionals. But it’s clear that GenAI offers solutions that can help
to bridge the gap between growing healthcare needs and apparently dwindling
healthcare resources. We needn’t be afraid of this. GenAI tools have demon-
strated remarkable medical knowledge – to the extent that some are competent
enough to pass medical certification exams.15 And as we’ve seen in this chapter,
they can even outperform human doctors when it comes to diagnosing issues.
So a future where clinicians and GenAIs work together – where human intui-
tion and expertise combines with machine precision – is enticing. As healthcare
systems become more stretched, combining human and machine expertise will
likely be the best way to diagnose patients and provide appropriate treatment.

What’s more, GenAI tools can empower individuals to take charge of their
own healthcare, by engaging in discussions with healthcare chatbots about
potential diagnoses and treatment options, and seeking medical help when
needed. Think of it as the new version of Googling your symptoms, only
much more advanced and precise! And when a patient does interact with a
doctor, GenAI can add value in the background, generating medical notes
and helping doctors search information.

But there are challenges

When a GenAI tool generates inaccurate or “hallucinated” content in a sec-
tor like marketing, it’s probably rather embarrassing for the company using
that content. But if a healthcare GenAI generates inaccurate or incoherent
responses, that’s a whole other story. So while large language models and

150
HEALTHCARE TRANSFORMATION: FROM PERSONALIZED ADVICE TO OPERATIONAL IMPROVEMENTS

other GenAI models hold great promise for solving problems in healthcare,
providers obviously need to ensure the accuracy and reliability of responses.

Are there challenges to deploying GenAI in healthcare? Absolutely. But given

the current state of healthcare provision and access around the world, I believe
GenAI promises more solutions than obstacles.

Key Takeaways
To summarize the key applications of GenAI in healthcare:

• Through more intelligent, responsive chatbots, GenAI can be used to

deliver personalized advice to patients, allowing them to discuss con-
cerns, get potential diagnoses, and understand when they may need
treatment. New tools are already emerging to meet this demand for
more immediate access to healthcare advice.
• When it comes to the early detection of diseases, AI has already proven
itself to be as capable and reliable as human doctors at interpreting
medical images and generating reports based on those images. Not
only can this help radiologists in their work, it can also help doctors
in other departments (such as the ER) to quickly diagnose issues and
accelerate patient treatment.
• Personalized treatment is a particularly hot area in medicine, but one
that’s challenging to deliver given each patient has a different medi-
cal history, different needs, and a different lifestyle. GenAI can help
clinicians overcome this challenge by acting as a medical “co-­pilot” –
­suggesting potential treatments and next steps.
• Behind the scenes, GenAI can also deliver huge operational i­ mprovements –
particularly when it comes to streamlining medical notetaking, dealing
with routine calls and enquiries at clinics, and more. This could improve
the patient experience and help to reduce clinician burnout.
• Healthcare systems around the world are struggling and waiting times
can be frustratingly long. GenAI can help to solve some of the biggest
challenges facing global healthcare systems and make healthcare much
more accessible and efficient.

151
 GENERATIVE AI IN PRACTICE

Now let’s turn to something completely different: the world of gaming. In the
next chapter, we’ll explore how video game design is being transformed by
GenAI, and how the games of the future will be more responsive, personal-
ized, and immersive than ever before. Ready Player One? Then turn the page.

Notes
1. About Ada; Ada; https://ada.com/about/
2. Is generative AI the future of rapid and accurate chest radiograph interpreta-
tion in the ER?; News Medical Life Sciences; https://www.news-­medical.net/
news/20231008/Is-­generative-­A I-­t he-­f uture-­of-­r apid-­and-­a ccurate-­chest-­
radiograph-­interpretation-­in-­the-­ER.aspx
3. Accuracy of a smartphone application for triage of skin lesions based on machine
learning algorithms; Journal of the European Academy of Dermatology & Venere-
ology; https://onlinelibrary.wiley.com/doi/10.1111/jdv.15935
4. New Generative AI-­Native Health Company RhythmX AI Announces Pre-
cision Care Platform For Doctors to Deliver Hyper-­ Personalized Care to
the Right Patient at the Right Time; PR Newswire; https://www.prnewswire.
com/news-­releases/new-­generative-­ai-­native-­health-­company-­rhythmx-­ai-­
announces-­precision-­care-­platform-­for-­doctors-­to-­deliver-­hyper-­personalized-­
care-­to-­the-­right-­patient-­at-­the-­right-­time-­301947898.html
5. Conversational AI for healthcare; Hyro; https://www.hyro.ai/healthcare/
6. NextGen Healthcare reveals AI notetaking product; Healthcare Dive; https://
www.healthcaredive.com/news/nextgen-­healthcare-­generative-­ai-­clinical-­
documentation-­ambient-­assist/696029/
7. Yet another study highlights EHR burden on physicians; Healthcare Dive;
https://www.healthcaredive.com/news/another-­study-­highlights-­ehr-­burden-­
on-­physicians/504805/
8. In pilot, generative AI expected to reduce clinical documentation time at Bap-
tist Health; Healthcare IT news; https://www.healthcareitnews.com/news/
generative-­ai-­reduces-­clinical-­documentation-­time-­baptist-­health
9. Google Cloud Launches New Generative AI Capabilities for Healthcare; Forbes;
https://www.forbes.com/sites/saibala/2023/10/09/google-­cloud-­launches-­new-­
healthcare-­generative-­ai-­features/?sh=3de773ac7e5b
10. An AI Tool That Can Help Forecast Viral Outbreaks; Harvard Medical School;
https://hms.harvard.edu/news/ai-­tool-­can-­help-­forecast-­viral-­outbreaks
11. Health Workforce; World Health Organization; https://www.who.int/health-­
topics/health-­workforce#tab=tab_1

152
HEALTHCARE TRANSFORMATION: FROM PERSONALIZED ADVICE TO OPERATIONAL IMPROVEMENTS

12. New AAMC Report Reinforces Mounting Physician Shortage; AAMC;

https://www.aamc.org/news/press-­r eleases/new-­a amc-­r eport-­c onfirms-­
growing-­physician-­shortage#:~:text=According%20to%20new%20data%20
published,and%20specialty%20care%2C%20by%202033
13. NHS waiting list in England rises to record 7.75 million; BBC News; https://
www.bbc.com/news/health-­67087906
14. Universal health coverage; World Health Organization; https://www.who.int/
news-­room/fact-­sheets/detail/universal-­health-­coverage-­(uhc)
15. Generative AI in Medicine and Healthcare; Promises, Opportunities, and
Challenges Future Internet; https://www.mdpi.com/1999-­5903/15/9/286

153
 12
VIDEO GAME DESIGN
AND TESTING:
THE GENERATIVE AI
APPROACH

I wanted to include a chapter on video games because it’s something that is a

little different and also since there are some really interesting advances hap-
pening in this area. Plus, I’m sure a lot of my readers are keen gamers and
would be interested to know how games will evolve in the GenAI era. (Hint:
we can look forward to more personalized experiences and NPCs [non-­player
characters] that are capable of proper conversations.)

GenAI will play such a key role in the video game industry that one survey
of video game execs found that GenAI is expected to contribute to more than
half of video game development within the next 5–10 years – freeing up game
designers from the more mundane development work and helping to reduce
costs and development time.1

In addition to helping develop new, better video games, GenAI also has a role
to play in quality control and testing of video games. Let’s explore both factors,
along with some emerging examples from the video game industry. (Along-
side, you can read about the impact of GenAI on coding in Chapter 16.)

155
 GENERATIVE AI IN PRACTICE

Generative AI in Video Game Development

and Testing
In this age of GenAI, creating video game environments and characters no
longer needs to be done painstakingly by hand. GenAI can automate many
development tasks, freeing up developers to focus their time on the most
impactful aspects of game design.

Plus, as GenAI tools become more sophisticated, it will be possible to build

the technology into real-­time games that offer personalized environments
and NPCs that adapt and evolve based on a player’s style and strategy – as
opposed to the pre-­built worlds and pre-­programmed NPCs of today’s video
games. This could potentially transform the gaming industry, ensuring each
gamer’s experience is different.

GenAI uses in game development

For developers, GenAI offers exciting new ways to create engaging content,
realistic visuals, and immersive gameplay experiences. As an example, let’s say
part of a video game takes place in a forest. Pay attention and you may start to
notice that there’s only a small number of individual tree models that repeat
over and over again. Once you’ve noticed it, you can’t help but keep seeing
the same trees, which is jarring and takes you out of the game experience.
But if that forest was created with GenAI, it could potentially be populated by
thousands of completely unique trees and critters – which is not only more
visually appealing and immersive, it keeps you focused on the gameplay expe-
rience, without being distracted by technical limitations.

Let’s break down some of the main elements of game development and see
how GenAI can facilitate the process:

• Procedural generation: By leveraging algorithms to create vast, intri-

cate, and unpredictable environments, games can potentially deliver a
unique, dynamic experience in each gameplay session.
• Terrain generation: GenAI can assist in the generation of realistic ter-
rains, enhancing the visual appeal of game worlds.

156
 VIDEO GAME DESIGN AND TESTING: THE GENERATIVE AI APPROACH

• Automated modeling: GenAI can streamline the creation of 3D

­models for characters and other elements, thereby reducing the time
and expertise required.
• Voice synthesis: Developers can utilize AI to create diverse and realis-
tic voiceovers for characters.
• Voice modification: GenAI provides the means to modify voice
recordings easily, thus creating a diverse range of voices for characters,
with limited resources.
• Generative music: Thanks to GenAI, it’s possible to automatically
create music scores for games – and even create dynamic scores that
change based on in-­game events, thereby enhancing the emotional
impact of games.
• Sound effects: GenAI can also be used to generate realistic sound
effects – again, potentially on-­the-­fly, to support the immersion and
realism of game worlds.
• Ambient soundscapes: Immersive ambient soundscapes that enhance
the game’s atmosphere can be generated with GenAI.
• Dialogue generation: GenAI can also write realistic dialogue for char-
acters, again, potentially on-­the-­fly.
• Personality generation: GenAI is capable of crafting unique and con-
sistent character personalities that enrich storytelling.
• Dynamic storytelling: Non-­linear storylines that adapt based on player
choices are another possibility thanks to GenAI. This helps to create a
more personalized gaming experience.
• Automated quest generation: GenAI can also be used to automati-
cally create quests and missions that align with the game’s overarching
narrative.

Bottom line, GenAI can autonomously create hugely immersive and engag-
ing gaming elements – thereby speeding up game development and reducing
the costs associated with bringing new games to market. And in the future,
GenAI will be increasingly used to create dynamic elements, characters, and

157
 GENERATIVE AI IN PRACTICE

worlds that can evolve and adapt to each player. Make no mistake, GenAI
marks a huge leap forward in video game development.

GenAI in video game testing

Another way GenAI can add value to video games is in game testing and
quality assurance. GenAI can be used to automatically detect bugs and defects
in games, and rank them according to priority and their impact on the game.
Plus, the predictive capabilities of AI mean it can be used to analyze data to
predict potential issues within a game, allowing developers to take proactive
action. GenAI can also be used to create multiple simulated players (i.e., bots)
to test the game, all playing in different ways according to their AI-­generated
play styles. This allows game developers to quickly work out where players get
stuck, where players are most likely to spend money (on in-­game accessories),
and which factors lead to less satisfying gaming experiences.

All of this will help to speed up the quality assurance process, while simulta-
neously increasing the accuracy and overall experience of video games.

Examples of GenAI in games

Let’s turn to some real-­world examples that show how developers are already
using GenAI in their work.

Video game company, Ninja Theory, has reportedly tapped into GenAI to
generate vocal performance – using the Altered AI voice library.2 Meanwhile,
Ubisoft, creator of Assassin’s Creed, has developed its own in-­house GenAI
tool called “Ghostwriter” to create “barks,” those brief phrases spoken by
NPCs when triggered by certain events.3 The idea is to automatically create
first drafts of character dialogues (e.g., enemy dialogue during a battle scene),
which scriptwriters can then select from and, if necessary, polish up. By auto-
mating these little lines of dialogue – of which there could be hundreds within
a game – scriptwriters can instead focus their time on core plot dialogue.

It’s no wonder the developer of Ubisoft’s Assassin Creed game, Jade Raymond,
says AI in the development of big-­budget games is “unavoidable” due to its
ability to cut expenses and speed up game development.4

158
 VIDEO GAME DESIGN AND TESTING: THE GENERATIVE AI APPROACH

Hexworks, creators of the game Lords of the Fallen, also used GenAI during
development – not to create artwork, dialogue, or characters, but to optimize
the game.5 Today’s stunning graphics-­heavy games present problems in terms
of memory usage and computational oomph (technical term). Hexworks says
GenAI can help with this by automatically detecting inefficiencies and sug-
gesting solutions, such as reducing texture sizes. Interestingly, AI-­generated
voices were also used in an early build of the game, with voice actors being
brought in for the final version. For me, this shows that GenAI can help to
speed up prototyping and get new game projects up and running faster, with-
out replacing human creativity.

It’s even possible to create entire games with GenAI. That’s the case with the
Angry Pumpkins game, inspired by the popular Angry Birds game and cre-
ated entirely using GenAI – with ChatGPT being used to write the code, and
AI imaging tools like Midjourney and Dall-­E creating the images. Its creator,
Javi Lopez, says the game is more of a “proof of concept” rather than a direct
competitor for Angry Birds. Basically, he wanted to show what was possible
with GenAI tools.6 And the end result is pretty impressive, especially consid-
ering its developer didn’t have to write a single piece of code. It just goes to
show how easy it is to create games with GenAI.

GenAI is also helping to breathe new life into classic computer games. Revolu-
tion Software, the UK game developer that scored a big hit with the Broken
Sword series of adventure games in the 1990s, has been using GenAI to help
update the early games in the series so they can be played on the latest genera-
tion of consoles and PCs. The first games in the series had hand-­drawn graph-
ics that were scaled to fit the far lower resolution displays that were used in the
1990s. Recreating all that hand-­drawn artwork at a higher resolution would
have been prohibitively expensive. So the team worked with GenAI research-
ers at the University of York, who were able to take a few sample pieces of art
designed for a modern-­day update and use them to train a GenAI model – the
result being a model capable of creating a piece of in-­game artwork (such as an
object or character) in as little as five minutes.7 Human artists were then used
to retouch some of the AI-­generated art (e.g., making faces more realistic). So
basically, instead of having to redraw everything, the team could focus human
efforts where it was most beneficial, and let GenAI take care of the rest. And
this made the whole endeavor economically feasible for Revolution Games.

159
 GENERATIVE AI IN PRACTICE

New GenAI tools for developers

Meanwhile, new tools are emerging to help developers harness GenAI. As an
example, leading game software company Unity has announced an AI market-
place where developers can gain access to GenAI solutions that help to accelerate
game development – including solutions for asset creation, NPC platforms and
voice generators.8 Unity has also created the Muse and Sentis generative plat-
forms, which further help developers harness AI technology in their workflows.

There’s also a new animation tool called “Animate Anything,” created by Any-
thing World, which can be used to easily animate 3D models. This means
users can upload a static 3D model and have the tool create an animated 3D
version in minutes.9

Or there’s the SyncDreamer GenAI tool, which creates multiple 2D perspec-

tives of an object based on a single 2D image (whether a drawn image or a
photorealistic image). This allows game developers to streamline the process
of creating 3D representations from 2D images – something that has proven
stubbornly difficult and laborious until now.10

Similarly, there’s the Auctoria platform, which uses GenAI to create video
game assets and game levels from scratch. Users can convert images into 3D
equivalents, or create assets using simple text prompts.11

Microsoft is also embracing GenAI, partnering with Inworld AI to create Xbox

tools that enable game developers to create GenAI characters, storylines, and
quests. Additionally, the “AI design co-­pilot” can create detailed scripts and
dialogue trees.12 Microsoft’s partner, Inworld, has also been working on using
GenAI to create NPCs that react to questions from a player.

Elsewhere, Google Cloud has formed strategic partnerships with the leading
game developers in Vietnam, enabling the use of Google Cloud GenAI solutions
for games. Plans include building advanced chatbots for more realistic conversa-
tions with characters, and enabling games that respond dynamically to players.13

In another example, software company NVIDIA has unveiled its Avatar Cloud
Engine (ACE) for Games, a custom GenAI model that “aims to transform

160
 VIDEO GAME DESIGN AND TESTING: THE GENERATIVE AI APPROACH

games by bringing intelligence to NPCs through AI-­powered natural lan-

guage interactions.”14 As NVIDIA points out, even in today’s most sophisti-
cated games, interactions with NPCs still tend to be transactional, scripted,
and brief, with dialogue options being quickly exhausted. But with ACE for
Games, developers can improve the conversational skills of NPCs, making
them more intelligent with consistent personalities that evolve over time.
Importantly, this enables dynamic responses that are unique to the player. So
instead of brief, scripted conversations, gamers could, in the future, have end-
less interesting conversations with NPCs.

Generative AI Will Shape the Future

of Video Games
GenAI will be so impactful in game development that Xbox CFO, Tim ­Stuart,
has described its power as “magical.” Speaking at the MIT Sloan Gaming
Conference in 2023, Stuart said GenAI will help even more people create
games, describing “a world where we go from 2 million or 3 million core game
developers to 200 million game developers that can now use AI as a tool.”15
He also noted that gaming business models are shifting – moving away from
the model where companies sell a console and games, to a model where gam-
ers have more flexible ways to access games. Which means key performance
indicators revolve less around sales of consoles and games, and more around
factors like the number of hours played and revenue per hour (e.g., from play-
ers buying accessories within the game). Which means it’s more important
than ever to keep players immersed in games for longer periods – something
that GenAI can support.

As we’ve seen in this chapter, GenAI is likely to bring huge changes to the
video game industry by automating some of the menial development and
testing tasks, and enabling the creation of truly dynamic games. Think about
it – developers will no longer have to painstakingly code every single line
of dialogue or every little movement by an NPC; a GenAI tool can do it for
them, even reacting on-­the-­fly to individual player interactions. If you’re a
gamer who likes to feel fully immersed in a world and its characters, this is no
doubt a tantalizing concept.

161
 GENERATIVE AI IN PRACTICE

In other words, while GenAI can help to unlock huge efficiencies in game
development workflows, perhaps the biggest win of all is for the game-­playing
experience. That’s why new game studio Jam & Tea Studios is putting GenAI
at the heart of its vision – because of its ability to “empower previously impos-
sible levels of player agency . . . While we’re confident there will be improve-
ments to workflows, we are far more focused on using AI to unlock game
experiences folks have never seen before.”16

Looking further ahead, we may even see tools emerge that allow players to
create their own games, without any development knowledge. Pahdo Labs,
the studio behind the Halcyon Zero anime game, is planning to launch its
own GenAI tools for game creation, after raising US$15 million in funding.17
The plan is to open up creator tools to players, allowing them to build their
own anime games with GenAI.

Does all this mean we’ll no longer need game developers? No. There will
always be a vital place for human creativity in game development. But as with
other industries, the work of developers will evolve as they increasingly work
alongside AIs to create games.

Key Takeaways
In this chapter, we’ve learned that the video game industry is set to be trans-
formed by GenAI. Key points to note are:

• GenAI marks a huge leap forward in video game development. The

technology can be used to create immersive environments, game assets,
levels, objects, music, dialogues, and NPCs that make for a more realis-
tic, immersive, and engaging gaming experience.
• In the future, GenAI will enable games to evolve dynamically based on
the individual player, with characters and the storyline reacting to the
player’s style and strategy.
• GenAI can also help to streamline game testing and quality assurance,
for example, by creating bot testers that play the game using multiple
different styles to spot flaws.

162
 VIDEO GAME DESIGN AND TESTING: THE GENERATIVE AI APPROACH

From the world of gaming to something completely different: the legal sphere.
Let’s find out how GenAI can assist in the work of lawyers – and potentially
make legal advice more affordable and accessible to anyone.

Notes
1. Generative AI will contribute to more than half of video game development
within next 5 to 10 years, finds Bain & Company; Bain & Company; https://
www.bain.com/about/media-­center/press-­releases/2023/generative-­ai-­will-­
contribute-­to-­more-­than-­half-­of-­video-­game-­development-­within-­next-­5-­to-­
10-­years-­finds-­bain-­-­company/
2. How generative AI is revolutionizing the gaming landscape; Techwire Asia;
https://techwireasia.com/2023/06/gaming-­a i-­revolutionizing-­g enerative-­
landscape-­for-­developers-­and-­gamers/
3. Ghostwriter AI tool; Ubisoft; https://news.ubisoft.com/en-­u s/article/
7Cm07zbBGy4Xml6WgYi25d/the-­c onvergence-­o f-­a i-­a nd-­c reativity-­
introducing-­ghostwriter
4. Assassin’s Creed Developer Says AI Is Unavoidable in Future AAA Titles; Fan-
dom Wire; https://fandomwire.com/assassins-­creed-­developer-­ai-­unavoidable-­
aaa-­titles/
5. Lords of the Fallen dev experimented with generative AI during development;
Creative Bloq; https://www.creativebloq.com/features/lords-­of-­t he-­f allen-­
created-­using-­gen-­AI
6. This Video Game Was Made Entirely Using ChatGPT 4 and AI Imaging Tools;
Augustman; https://www.augustman.com/my/gear/gaming/chatgpt-­game-­like-­
angry-­birds-­made-­using-­ai-­tools/
7. Generative AI Is Breathing New Life into Classic Computer Games; Forbes;
https://www.forbes.com/sites/bernardmarr/2023/09/22/generative-­a i-­i s-­
breathing-­new-­life-­into-­classic-­computer-­games/
8. Generative AI Will Change Game Dev Forever; Analytics India; https://
analyticsindiamag.com/generative-­ai-­will-­change-­game-­dev-­forever/
9. New AI tool could be a game-­changer for 3D animation; Creative Bloq; https://
www.creativebloq.com/news/anything-­world-­animate-­anything-­ai
10. This generative AI model can transform the gaming industry; Cryptopolitan;
https://www.cryptopolitan.com/generative-­ai-­can-­transform-­gaming-­industry/
11. Auctoria uses generative AI to create video game models; TechCrunch; https://
techcrunch.com/2023/09/20/auctoria-­u ses-­generative-­ai-­to-­c reate-­v ideo-­
game-­models/
12. Microsoft is bringing AI characters to Xbox; The Verge; https://www.theverge.
com/2023/11/6/23948454/microsoft-­x box-­generative-­ai-­d eveloper-­t ools-
­inworld-­partnership

163
 GENERATIVE AI IN PRACTICE

13. Google Cloud forms strategic partnerships with Vietnam’s top game devs; Data
Center News; https://datacenternews.asia/story/google-­cloud-­forms-­strategic-­
partnerships-­with-­vietnam-­s-­top-­game-­devs
14. Generative AI Sparks Life into Virtual Characters with NVIDIA ACE for Games;
NVIDIA; https://developer.nvidia.com/blog/generative-­ai-­sparks-­life-­into-­
virtual-­characters-­with-­ace-­for-­games/
15. Xbox CFO on gaming content, business models, and generative AI; MIT; https://
mitsloan.mit.edu/ideas-­made-­to-­matter/xbox-­cfo-­gaming-­content-­business-­
models-­and-­generative-­ai
16. Industry vets form Jam & Tea Studios to build multiplayer RPG using generative
AI; Game Developer; https://www.gamedeveloper.com/business/industry-­vets-­
form-­jam-­tea-­studios-­to-­build-­multiplayer-­rpg-­using-­generative-­ai
17. Pahdo Labs Raises $15M to Let Players Make Anime Games with AI Tools; Decrypt;
https://decrypt.co/155895/pahdo-­labs-­raises-­15-­million-­players-­make-­anime-
­games-­ai-­tools

164
 13
THE LEGAL SPHERE:
AI-­ASSISTED DOCUMENT
CREATION AND REVIEW

You may raise an eyebrow at AI being used in the legal sector. But now, thanks
to the latest large language models, GenAI is much better at understanding
the context and nuance of complex legal texts. As such, it is proving a useful
tool in the work that lawyers do, particularly when it comes to reviewing legal
documents and drafting text.

GenAI has so much promise in the legal sphere that two-­thirds of large
law firms in the United Kingdom say they are exploring the potential of
GenAI (along with approximately a third of smaller law firms).1 Meanwhile,
in the United States, 67% of in-­house counsels expect their outside law
firms to use GenAI tools in their work.2 Another report asked 379 lawyers
and other legal professionals about ChatGPT – with more than 90% saying
they’d used it.3 We can therefore expect the workload of lawyers and other
legal professionals to change as they increasingly find themselves working
alongside AIs.

But that’s not all. As we’ll see in this chapter, GenAI can also help everyday
people tackle legal questions and issues.

165
 GENERATIVE AI IN PRACTICE

The Use of Generative AI in Law Firms

GenAI holds transformative potential for the legal profession. As GenAI sys-
tems become more sophisticated, they will facilitate the work of lawyers in
several ways.

Potential GenAI uses

We’ll get to specific examples later in the chapter. But for now, let’s explore over-
arching use cases. GenAI can support legal professionals with the following:

• Document drafting: GenAI can assist in automating the creation of

standard legal documents such as contracts, wills, and leases – tailoring
them to the client’s needs, according to specific parameters. GenAI can
also be used to draft communications to clients.
• Legal research: Researching existing case law and legal precedents can
be enormously time consuming. By providing a set of criteria or key-
words, lawyers can use GenAI to generate summaries or briefs on rel-
evant case law, statutes, and regulations.
• Discovery and data review: GenAI can assist in the discovery process
by generating queries to identify relevant documents or data. It can also
help in summarizing vast amounts of data.
• Contract analysis: AI can generate summaries, highlight key terms,
and identify potential issues in lengthy contracts.
• Predictive analysis: AI may even be used to predict the outcomes of
legal cases based on historical data, which can be useful for advising
clients or strategizing for litigation.
• Legal advice chatbots: For firms looking to expand their service
offering, GenAI can be used to create legal advice chatbots, which
generate responses to common legal queries on websites or through
messaging apps.

Such uses promise to enhance the efficiency of law firms, enabling lawyers
to tackle more cases in a shorter amount of time, while also reducing the

166
 THE LEGAL SPHERE: AI-­A SSISTED DOCUMENT CREATION AND REVIEW

margin for human error. Furthermore, AI can help democratize legal services
by making basic legal assistance more accessible and affordable to the general
public through automated platforms – but we’ll talk more about that later in
the chapter.

Here comes the caveat. . .

With these exciting advancements come challenges. Lawyers will need to
adapt their roles, acquiring new skills to effectively utilize and supervise these
technologies – for example, making sure the GenAI is delivering accurate out-
puts. There are also ethical considerations, particularly when it comes to the
potential for over-­reliance on GenAI, which could lead to a lack of nuanced
human judgment that is so crucial in legal situations. There may also be issues
around data bias that could skew GenAI responses. (A major concern for any
law firm, but especially in the case of, say, a legal chatbot delivering biased
advice and answers.) And there’s the risk that information given to a GenAI
tool by a law firm could be used for AI training purposes and potentially dis-
closed to other users.

In essence, while generative AI presents an array of opportunities for the legal

world, its integration must be approached thoughtfully to ensure that the
essence and integrity of the legal profession remain intact. We’ll talk more
about GenAI implementation in Chapter 18.

Examples from the legal profession

Now let’s look at some examples of how law firms and other legal organiza-
tions are beginning to harness GenAI.

You may have heard of the California Innocence Project (CIP), which was
established by the California Western School of Law to assist in cases where
there is compelling evidence that someone has been wrongly convicted. This
is complex, labor-­intensive work requiring many hours of reviewing case
law and poring over evidence for each case. So CIP has turned to GenAI to
help speed up the painstaking work of its lawyers – and hopefully, reduce
the amount of time innocent people spend behind bars. CIP lawyers have

167
 GENERATIVE AI IN PRACTICE

deployed a GenAI platform called “CoCounsel,” which was developed by AI

company, Casetext, in partnership with OpenAI. CoCounsel, which has been
specifically trained on legal documents, case law, and court proceedings, is
capable of reviewing and summarizing legal documents, drafting documents,
analyzing contracts, and preparing depositions. According to CIP lawyers,
using the tool for essential but mundane tasks saves precious time – for
example, reducing the time spent drafting a letter to a client from 15 minutes
down to one minute – thereby allowing the team to spend their time on more
impactful tasks that require human interaction.4 Importantly, in order to
reduce the potential for bias and “hallucinations” (i.e., fabricated responses),
CoCounsel was beta-­tested by 400 lawyers from boutique and multinational
firms – and Casetext says the sensitive information entered into the system
isn’t stored or fed back into the platform for training.5

Dentons, one of the world’s largest global law firms, is launching its own pro-
prietary ChatGPT-­style tool called “fleetAI,” based on the GPT-­4 large language
model. Dentons says the tool will enable its lawyers to “conduct legal research,
generate legal content, and identify relevant legal arguments.”6 Again, Dentons
says all data uploaded won’t be used to train the model, can’t be accessed by any-
one outside of the firm, and is deleted after 30 days. Plus, any Dentons lawyers
using the tool are required to independently verify and validate all outputs, and
disclose to clients that they have interacted with the AI tool as part of their work.

As you might expect, there is a plethora of GenAI-­powered tools and services

springing up to serve legal firms. As an example, legal research and software
company LexisNexis is embracing GenAI, building the technology into its
Lexis+ AI platform – which is designed to help lawyers conduct case research,
and draft summaries, legal documents, and communications.7 With a sim-
ple interface that acts rather like ChatGPT (indeed, the platform is built on
­OpenAI’s GPT language model, among other models), the GenAI responds
to prompts in seconds. It also provides citations for its responses. And as with
Casetext’s CoCounsel platform, Lexis+ AI is focused on private interactions,
to ensure that sensitive information won’t be shared with other users.

Another GenAI service designed for the legal profession comes from AI start-
­up Harvey, which builds custom large language models for law firms. Among
the firms partnering with Harvey are major London law firm Macfarlanes

168
 THE LEGAL SPHERE: AI-­A SSISTED DOCUMENT CREATION AND REVIEW

and multinational firm Allen & Overy.8 Lawyers can use Harvey for conduct-
ing research, analyzing documents, summarizing texts, and creating draft
communications (for human lawyers to review).

Information specialist Thomson Reuters is also building GenAI into its legal
offerings, including the Westlaw legal research platform and Practical Law,
which provides how-­to guides, templates, and checklists.9 GenAI means law-
yers and legal researchers can engage with these platforms in a more intui-
tive, conversational way and get natural language answers to their questions.
Thomson Reuters also offers GenAI-­powered contract review software called
“Document Intelligence,” which it says can cut the amount of time spent
searching for information in contracts by 50%.10

Helping Everyday People Digest Legal

Documents and More
Clearly GenAI can help cut the workload of lawyers, which in itself may help
to make legal advice more affordable and accessible for people. But, we will
also see new tools become available that allow lay people to directly interact
with GenAI legal tools. This can prove useful in several ways, such as digest-
ing lengthy contracts and legal letters, drafting contracts, and even accessing
basic legal advice. It’s early days in this field, but some interesting tools are
beginning to emerge that promise to democratize legal advice.

One example is called “Legal Robot,” which is designed to automatically

translate “legalese” into everyday language that anyone can understand. So
a lay person faced with a complex contract or other legal document can get
a plain-­English translation and properly understand what they’re reading or
signing. (Legal Robot also says its tool can be used by lawyers to ensure their
documents are easy for anyone to understand, but whether lawyers actually
want their documents to be easily understood is another matter!)

Another example comes from AI Lawyer, which offers 24/7 expert legal help
and information – so, you can, in theory, ask any legal question and get a
response. You can also use the tool to draft contracts without having to hire a
lawyer. Plus, it can be used to simplify or summarize legal text.

169
 GENERATIVE AI IN PRACTICE

It’s really exciting seeing tools emerge to help people access legal support with-
out the huge bills that can come with engaging a law firm. Who knows, maybe
developments like this will further encourage legal firms to embrace GenAI
in order to provide a more streamlined, affordable service to the public.

Key Takeaways
In this chapter, we’ve learned that GenAI is impacting the legal sector in two
main ways:

• Legal firms are embracing GenAI to help lawyers work in a more effi-
cient way. In particular, GenAI can be used to review legal documents
and cases, analyze contracts, draft documents, prepare depositions, and
write communications. Several GenAI platforms and tools are emerg-
ing to serve the legal profession.
• GenAI can also help to make legal advice more accessible to everyday
people, thanks to new tools that offer simple legal advice, and help with
translating complex legal jargon into everyday language. Such tools can
even help people draw up contracts without having to hire a lawyer.

Now let’s turn to a topic that spans many sectors and industries: design and
innovation. In the next chapter, we’ll see how GenAI is helping to enhance
the research and design process in everything from fashion and jewelry to
potentially life-­saving drugs.

Notes
1. Two-­thirds of large law firms researching generative AI; Legal Futures; https://
www.legalfutures.co.uk/latest-­news/two-­thirds-­of-­large-­law-­firms-­researching-­
generative-­ai
2. 60% of in-­house counsel expect law firms to use generative AI; Legal Dive;
https://www.legaldive.com/news/law-­firms-­generative-­ai-­use-­cases-­lexisnexis-­
ai-­survey/691913/
3. Generative AI: a legal revolution is coming – eventually; FT; https://www.ft.com/
content/0f36eb4e-­b90f-­4ffe-­befc-­daf01829c182

170
 THE LEGAL SPHERE: AI-­A SSISTED DOCUMENT CREATION AND REVIEW

4. How Generative AI Is Used to Fight Miscarriages of Justice at the California Inno-

cence Project; Forbes; https://www.forbes.com/sites/bernardmarr/2023/10/06/
how-­generative-­ai-­is-­used-­to-­fight-­miscarriages-­of-­justice-­at-­the-­california-­
innocence-­project/
5. Casetext Unveils CoCounsel, the Groundbreaking AI Legal Assistant Powered
by OpenAI Technology; PR Newswire; https://www.prnewswire.com/news-­
releases/casetext-­unveils-­cocounsel-­t he-­g roundbreaking-­ai-­legal-­assistant-­
powered-­by-­openai-­technology-­301759255.html
6. Dentons to launch client secure version of ChatGPT; Dentons; https://www
.dentons.com/en/about-­dentons/news-­events-­and-­awards/news/2023/august/
dentons-­to-­launch-­client-­secure-­version-­of-­chatgpt
7. LexisNexis announces new generative artificial intelligence platform; ABA
Journal; https://www.abajournal.com/web/article/lexisnexis-­announces-­new-­
generative-­ai-­platform-­lexis-­ai
8. UK law firm is latest to partner with legal AI startup Harvey; Reuters; https://
www.reuters.com/legal/transactional/uk-­law-­firm-­is-­latest-­partner-­with-­legal-­
ai-­startup-­harvey-­2023-­09-­21/
9. AI @ Thomson Reuters; Thomson Reuters; https://www.thomsonreuters.com/
en/artificial-­intelligence.html
10. Document Intelligence; Thomson Reuters; https://legal.thomsonreuters.com/en/
products/document-­intelligence

171
 14
CRAFTING TOMORROW:
AI IN DESIGN AND
DEVELOPMENT

GenAI is poised to revolutionize the design, research, and product

­development processes by injecting advanced computational power and crea-
tivity into these traditionally human-­centric domains. Drug development is a
super-­interesting and leading-­edge application of GenAI, so this chapter will
kick off with a look at how GenAI will shape the drug therapies of the future.
We’ll then explore the concept of generative design, and see how GenAI can
help design all manner of products. And we’ll finish up with a brief look at
digital twins and academic research. There’s a lot to cover in this chapter, so
let’s get started.

Generative AI in Drug Discovery

For me, drug discovery is an area that perfectly demonstrates the trans-
formative power of GenAI. The process of bringing a new drug to mar-
ket is long and eye-­wateringly expensive. It can take up to 15 years from
starting the discovery process to the point at which a drug is approved for
sale – and 9 out of 10 drugs that enter clinical trials never make it to the
approval stage. No wonder the cost of bringing a new drug to market can
reach US$2.5 billion, once you account for the cost of both successful and
failed programs.1 GenAI may help to speed up the drug discovery phase,

173
 GENERATIVE AI IN PRACTICE

thereby helping to get drugs to the clinical trial phase more quickly (and
with less money spent).

How GenAI is accelerating the search for new drugs

To get an overview of GenAI’s role in drug discovery, I spoke with Kimberly
Powell, VP and general manager of healthcare at NVIDIA.2 Powell’s team
works with industry leaders, academics, pharma companies, and tech bio
companies to apply AI to the drug discovery process. Essentially, they exist
to make AI technology accessible across the ecosystem. So if anyone knows
how GenAI could potentially transform the search for new drug therapies,
it’s Powell.

As Powell explained it to me, GenAI excels at bringing the world’s knowledge

together in one place. If we think of ChatGPT, for instance, it was trained
on everything available on the internet, including books and articles. This
natural language prowess can be applied to the R&D phase of drug discovery
by trawling through research papers and electronic health record data to dis-
cover, for example, how existing drugs are affecting people.

But what gets the team at NVIDIA really excited is the potential to apply
GenAI to other languages related to drug development: the languages of
biology and chemistry. Think of human DNA as a sequence of four let-
ters (A, T, C, and G) strung together into a 3-­billion-­letter-long sentence.
That’s a language of its own. We also have proteins, which form the building
blocks of biology. Proteins have their own alphabet – 20 letters for amino
acids, strung together in lengths of tens of thousands, or even hundreds of
thousands. Chemicals, too, have a language known as “SMILES” (Simplified
Molecular Input Line Entry System) – characters that, together, define the
structures of chemistry.

As Powell puts it, “We can now take these languages . . . and we can apply the
method of generative AI and GPT-­type methods . . . Once we do that, the
language models can really help us understand a lot more about biology that
maybe we haven’t been able to observe in the real world.” This means, not
only can we discover new drugs with GenAI, we can also do it with less time

174
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

and less cost. And in the process, Powell’s hope is we can improve that 90%
failure rate I mentioned earlier.

But how exactly does GenAI apply to the drug discovery process? Powell
described the drug discovery process as having three phases:

The first phase is about establishing the target (disease or condition) that they
want to go after with a new drug. In this phase, GenAI can be used to study
genomics, understand the gene that’s causing the disease, or understand what-
ever is happening in the body. Basically, the aim is to understand the target
better. In Chapter 11 we saw how one GenAI model was able to predict variants
of the COVID-­19 virus. Another example comes from the GenSLMs model,
developed by researchers from NVIDIA, Argonne National Laboratory, the
University of Chicago, and more. The model was trained on COVID-­19 genome
data from the first year of the pandemic, and from that it successfully generated
sequences that closely matched the Eris and Pirola variants from 2023.3 In other
words, it was able to accurately predict how the virus would later mutate. This
sort of modeling could potentially inform drug discovery in future, and would
also help governments track and manage infectious diseases.

The second phase of drug discovery is about coming up with leads, i.e., chem-
icals or proteins that could be used to target that disease. And this is where the
scale of the problem becomes truly mind-­boggling. Because there are more
than 1060 chemicals and 10160 proteins that could potentially be made to tar-
get a disease. No wonder drug discovery is often described as searching for
a needle in a haystack! GenAI can sift through these potential chemicals and
proteins and start generating ideas – potentially even inventing new chemi-
cals and proteins with the desired structure and function to target the disease
in question. This creates a tremendous number of new leads to explore, which
is really exciting.

And the third phase is about optimization. Say the GenAI model has generated
one billion compounds that could potentially be effective; the drug company
then needs to test those against the target. GenAI can assist with this screen-
ing process at a scale and speed that’s never been seen before. In one exam-
ple, NVIDIA worked on a project with Recursion Pharmaceuticals to screen
more than 2.8 quadrillion small-molecule-­ target pairs (using Recursion’s

175
 GENERATIVE AI IN PRACTICE

matchmaker GenAI algorithm running on NVIDIAs supercomputing tech-

nology). Within a week, they were able to complete screening that would have
taken 100 000 years with traditional methods.4

In short, then, GenAI can help pharma companies explore potential new
drugs with unprecedented scale, speed, and accuracy – which, in turn, allows
them to proceed to clinical trials more quickly. As an example of this, Insilico
Medicine used GenAI in the drug discovery process to develop a drug to
treat idiopathic pulmonary fibrosis. Using traditional methods, the process
would have cost more than US$400 million and taken up to six years. But
with GenAI, NVIDIA says Insilico accomplished the task in one-­tenth of the
cost and one-­third of the time, proceeding to clinical trials in just two and a
half years.5 (Incidentally, Insilico has also made the news for developing a new
AI-­generated COVID drug that has entered clinical trials, and is reported to
be effective against all variants.)6

Imagine that sort of acceleration across the entire pharma industry and you
can see how GenAI can absolutely transform this aspect of healthcare.

Enabling truly personalized medicine

Looking further ahead, Powell says GenAI will also help to realize truly per-
sonalized medicine. GenAI can be used to analyze patient genetic and health
data to suggest personalized drug treatments based on the individual’s unique
profile. This approach might be particularly beneficial in cancer therapy,
where the genetic makeup of tumors can vary widely.

As Powell predicts, “This true opportunity of personalized medicine will only

be realized through these generative methods. Because you have to associate
so much information: my history, my DNA, what you’ve discovered about
whatever disease state I have . . . it’s completely unique to me. And only
through these methods and being able to synthesize that level of information
will we get there. And once you understand my issue, then you can have a
very fast-­track discovery process so that you can treat me, not in the five years
it takes to discover a drug, but maybe in five weeks.”

176
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

Other examples from the world of medicine

Of course, NVIDIA isn’t the only provider working to aid drug discovery.
Pharma company Merck has developed a GenAI-­powered drug discovery
platform called “Addison,” which is designed to help drug designers explore
that vast chemical landscape I mentioned and come up with drug candidates
much more quickly.

As Merck explains, there are an estimated 166 billion molecules in the chemi-
cal universe database GDB-­17, yet even libraries that vast “barely scratch the
surface of all possible structures.”7 Which is where a platform like Addison
can add tremendous value – by helping researchers “explore unbounded
chemical space and generate ideas for entirely new compounds.”

In another example, GenAI was used by British biotech company Etcembly to

design a cutting-­edge immunotherapy drug for cancer. Called “ETC-­101,” the
new drug is a type of T-­cell engager (T-­cells being a type of white blood cell).
These types of drugs basically harness the power of T-­cell receptors (TCRs),
which recognize foreign antigens in the body, to identify and destroy cancer
cells. The trouble is, they can also recognize targets in healthy cells. Etcembly’s
solution to this is to engineer TCRs that are more sensitive and selective – so
they can effectively target cancer cells, with minimal side effects.

To do this, Etcembly deployed GenAI to decode the “language” of TCRs and

“rewrite” the genetic code of the TCR to make it more potent. The result-
ing drug, ETC-­101, has been designed to target PRAME, a protein associated
with many cancers known for their low survival rates. The drug reportedly
binds to PRAME with a millionfold greater affinity than natural TCRs (again,
without targeting health cells), and was developed in 11 months, as opposed
to the two-­year timeframe typical for TCR discovery.8 The drug may enter
clinical trials as early as 2025.

Elsewhere, Dell has teamed up with the University of Limerick to create an

AI platform for cancer diagnostics and predictive research. The goal is to
use AI to speed up cancer biomarker testing and enhance the treatment of
patients with B-­cell lymphoma.9

177
 GENERATIVE AI IN PRACTICE

GenAI may also play a role in artificial DNA synthesis (gene synthesis), which
allows virtually any DNA sequence to be generated in a laboratory setting.
Again, if we think of DNA as a language, it makes sense that GenAI can help
researchers “decode” and “rewrite” this language. Given that the majority of
cancers are caused by sporadic mutations of genes, as opposed to genes that
we’re born with, it’s clear that gene synthesis can help us understand more
about, and better treat, cancer. The human genome is enormously complex
and the array of genetic factors that contribute to health and disease is vast.
This poses a problem for researchers looking to understand disease. But tech-
nology – specifically GenAI – can help.

Google DeepMind is another tech company looking to support health and drug
research. It has created a new AI tool that classifies the effects of 71 million “mis-
sense” mutations – genetic mutations that result in changes in a DNA sequence.
These mutations can be benign or they can be pathogenic, and the average per-
son carries more than 9000 missense variants. Understanding these millions of
variants – particularly whether they are benign or disease-­causing – is a com-
plicated, resource-­heavy task. According to DeepMind, only 0.1% of all 71 mil-
lion possible variants have been classified as pathogenic or benign.10 DeepMind’s
AlphaMissense platform has been developed to help accelerate classification.

All of this is super-­exciting to me. But could it also pose a threat to humanity?
After all, if you can use AI to design new drugs, the other end of the spectrum
is using GenAI to design the most poisonous drugs or identify new toxins
that are dangerous to humans. (Case in point: AI was able to come up with
40 000 potential new chemical weapons in just six hours!)11 In the wrong
hands, this technology is terrifying. But the rewards of being able to better
understand and target diseases, and increase health spans of people around
the world, are worth striving for.

Designing Anything with Generative AI and

Generative Design
The field of generative design uses AI to conceptualize myriad design possibil-
ities that adhere to specific constraints and performance parameters, offering
a breadth of optimized solutions that would be nearly impossible for humans

178
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

to create manually. So, designers enter parameters such as what materials

should be used, the size and weight of the desired product, what manufactur-
ing methods will be used, and how much it should cost, and then the genera-
tive design algorithms cough up designs. This transformative approach not
only accelerates the ideation phase but also enhances innovation, allowing
designers and engineers to explore a wider range of design options, and then
pick and choose the best ones. And that applies to designing anything – prod-
ucts, buildings, potential solutions for climate change, anything at all.

Generative design software has been around for a few years, but now, we’re
beginning to see generative design being combined with GenAI to make the
process of designing with AI more intelligent and more intuitive. Indeed, the
rise of GenAI means you won’t necessarily need to be a professional designer
or an expert in design software to design amazing products in future. You’ll
be able to tell a GenAI interface what you want to achieve and it will come up
with various designs for you.

How GenAI will enhance generative design

To learn how GenAI is transforming the design process, I spoke with Mike
Haley, head of research at Autodesk. Autodesk has a huge focus on cutting-­
edge technologies – so much so that they were talking about generative design
as early as 2009.

Haley says that across all sorts of industries, Autodesk clients are trying
to design more complex things, with more complex requirements and
constraints, while at the same time wanting to design more efficiently and
cheaply. There’s a lot of pressure on the design process, basically, which is why
generative design is so valuable – it helps designers come up with complex
designs in a more efficient, cost-­effective way.

So how can GenAI make generative design even better? It turns out, genera-
tive design tools can actually be quite difficult to use because, to get the best
out of generative design software, you have to be able to specify what you want
in a very exacting way. Trouble is, most designers don’t n ­ ecessarily know all
the constraints right at the beginning of a design project. Often, constraints

179
 GENERATIVE AI IN PRACTICE

and requirements will emerge during the design process. Plus, generative
design solutions can be expensive because there’s a lot of computational
power involved.

GenAI can help to solve these issues. GenAI is fast, and it’s super-­easy and
intuitive to use because you can just talk to the system or type in what you
want. So, combining GenAI with generative design gives professional design-
ers (and even non-­professional designers) the best of both worlds. With a
generative design system, you get incredible precision, and with GenAI you
get a system that’s much easier to use. This is what Autodesk is particularly
excited about – combining the two technologies to make generative design
more accessible to more people, and to enhance the creativity of Autodesk
customers.

Haley’s Autodesk research lab works with forward-­thinking customers to

come up with new design solutions. He gave an example of working with one
client, a large construction company, who had concerns about how long it
took to do the early conceptual designs for skyscrapers. The customer found
this early explorative process took months, and they wanted to accelerate the
process for future buildings. By applying GenAI to this initial conceptual
design process, they were able to shrink the process from multiple months
down to a week. Clearly, GenAI’s ability to quickly synthesize and iterate on
design concepts helps designers explore a larger landscape of potential solu-
tions in a much shorter time period – leading to more innovative and creative
outcomes. Imagine such technology being applied across the design process
for pretty much everything – houses, products, cars, etc. – and you can see
how GenAI can add serious value to the design process.

(As an aside, Autodesk’s 60 000-­square-­foot Toronto offices, named “Autodesk

Technology Centre,” was one of the first AI-­designed offices of its scale. The
designers consulted with Autodesk employees about their preferences for fac-
tors like daylight, views, and proximity to other desks, and those parameters
were fed into the AI design tool.12)

What’s more, the combination of GenAI, generative design and 3D printing

may prove to be particularly impactful. Generative design and GenAI can
be used to better understand what can be done with 3D printing, since the

180
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

range and complexity of shapes that can be created with 3D printing is so vast.
Increasingly, this winning combination will be used to create new compo-
nents and products that are cheaper, lighter, and sturdier, thereby improving
the overall quality of many products, from cars and aircraft to prefabricated
houses and structures.

When I asked Haley about the future of GenAI and design, he was passionate
about the idea of design software (indeed, all software) becoming much easier
to use, because GenAI allows us to express ourselves however we want – by
speaking, typing, sketching, whatever. “The nature of user interfaces of all
software, whether it’s a word processor or a complex building design product,
are going to fundamentally change in the next five years . . . all of a sudden,
computers have this ability to interpret us.”

He also predicts that design software is going to become more proactive, inter-
jecting with helpful suggestions based on what it knows the user wants to design,
and what the system has learned from past data. “We’re going to see, emerging
in the design process, these tools that are not just overwhelming you with the
volume of data but are coming to say, ‘Hey, I notice you’re designing this. Here’s
something you may want to know that I’ve noticed in all the data that may be
relevant to what you’re designing now.’ So it helps you focus your attention.”

The business argument for GenAI-­powered design

(and manufacturing)
While many designers are already accustomed to designing with AI, it’s clear
that GenAI is going to further aid and improve the design process. It will give
designers the ability to generate designs more quickly, and work with AI in
a more intuitive way. It can be used to model how materials or products will
perform, without always having to build physical prototypes. It can even be
used to incorporate customer feedback data or user data from existing prod-
ucts (or buildings) into future designs. So, the whole design phase becomes
less of a trial-­and-­error process, basically; which clearly reduces the time and
resources involved. Plus, it can help to create better products. As an example,
with Autodesk’s help, General Motors was able to design a new car compo-
nent that was 40% lighter and 20% stronger than its predecessor.13

181
 GENERATIVE AI IN PRACTICE

All this will bring significant advantages to organizations, not least the abil-
ity to quickly innovate, explore a wide range of creative options in a shorter
time period, accelerate the process of iterating and refining prototypes,
and bring new (better) products to market, faster. Plus, GenAI opens up a
whole new world of product personalization, especially when combined with
3D printing.

GenAI can also be used as part of intellectual property processes. For exam-
ple, it can analyze whether patents are necessary in order to comply with
a particular product standard. It can analyze existing patent text, research
papers, and market trends to help companies target their R&D efforts. And it
can analyze vast amounts of data online to monitor for potential trademark
infringements.

We can also expect GenAI to become integrated into the manufacturing pro-
cess, which is why Microsoft and Siemens have partnered on a new GenAI assis-
tant designed specifically for manufacturing.14 The companies say the Siemens
Industrial Copilot tool will improve human-­machine collaboration, thereby
increasing productivity and speeding up manufacturing tasks. As an example,
the companies say tasks such as debugging automation codes and running sim-
ulations can be done in mere minutes using the GenAI assistant. Microsoft and
Siemens have plans to develop GenAI co-­pilot tools for a wide range of indus-
tries beyond manufacturing, including transportation and healthcare.

Examples from jewelry and fashion

Jewelry designers are already harnessing AI for their designs, enabling them
to push the boundaries of new designs, accelerate prototyping, and stream-
line the design process. Jewelry companies that have used AI as a co-­creator
include high-­end Parisian jeweler Boucheron, and Vancouver-­based private
jeweler Volund Jewelry.15 Combined with 3D printing, AI-­aided design allows
jewelers to quickly create prototypes and see how their designs look in the
real world (since printing even an elaborate prototype can be done in hours,
versus the weeks it might take to hand-­craft a wax mold). Tiffany & Company
is one major jewelry designer that has used both AI-­enhanced design and
3D printing to create prototypes.16

182
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

In the future, GenAI could even enable customers to design their own
bespoke jewelry. Using GenAI tools, customers could create their own unique
designs according to their preferred materials, design preferences, and cost,
and then generate images to see how the design would look on themselves
before they order. And with 3D printing – which can be used to print all sorts
of materials, including plastic, metals, concrete, and even chocolate – one-­
of-­a-­kind products could be manufactured without having to worry about
economies of scale.

Big-­name fashion and sports brands are also routinely using AI to augment
their design process, including Nike.17 It makes sense that fashion designers
can benefit hugely from AI-­enhanced design, since it enables them to origi-
nate multiple new designs more quickly.

In one example, fashion technology company, Fashable, has partnered with

Microsoft to bring GenAI to the clothing design process. Fashion is one
industry with particularly high expectations of quick turnarounds and end-
less new styles – driven especially by “fast fashion.” But instead of adding to
the ­problem of the “throwaway culture,” Fashable believes GenAI can help
designers meet demand while reducing waste. How? In part by allowing
designers to innovate multiple new designs without wasting fabric, but also by
using AI to remove much of the guesswork that goes into creating new cloth-
ing. After all, trends are unpredictable and it’s not obvious whether inventory
will sell. So Fashable uses AI to analyze data from multiple sources, including
social media and retail sites, to better understand trends. Designs can be aug-
mented in real time to take account of trends, and the resulting images can be
shared to social media for A/B testing with consumers. So designers can bet-
ter understand demand before going into production. According to F ­ ashable,
instead of taking months to get a new collection from design to store, it can
now be done in minutes – by designing quickly and marketing directly to
customers, without wasting fabric.18

Examples from the automotive industry

The automotive industry has been using AI to help design cars and compo-
nents for years. But GenAI can help to further streamline and improve the

183
 GENERATIVE AI IN PRACTICE

design process in several ways. For example, GenAI can be used to ­generate
new design concepts and virtual prototypes. The process of moving from
early concept ideas and sketches, through to the development of full-­sized
clay prototypes, can take many months – but GenAI can help to speed up the
early parts of this process by instantly translating 2D sketches into lifelike 3D
virtual models, enabling designers to explore a wide range of options in less
time. Designers can also use text prompts such as “rugged” or “sleek” to gen-
erate design ideas. Plus, for interiors, GenAI can be used to generate realistic
visuals of textures based on simple descriptions.

Toyota’s Research Institute has been experimenting with using GenAI

to speed up vehicle design. It has developed a new technique that allows
designers to integrate initial design sketches and engineering constraints
(around performance, safety, and usability) into GenAI tools early in the
creative process. According to Toyota, this approach could “reduce the
number of iterations needed to reconcile design and engineering consider-
ations.”19 In other words, this helps designers come up with designs accord-
ing to their creative vision (e.g., for a rugged, SUV-­type vehicle), while
optimizing for the many complex engineering and safety considerations
that are essential.

Speaking of safety, GenAI can also be used to detect and prevent defects in
vehicles, and generate realistic simulations of accidents, which can be used
to train autonomous vehicles and identify potential safety hazards in exist-
ing vehicles.

GenAI is even being used to help battery development for electric vehicles
(EVs). This is the focus of start-­up Aionics, which is using GenAI to speed
up EV battery research. It’s like the drug discovery process I talked about ear-
lier in the chapter – there are billions of procurable molecules that could be
combined to make electrolyte materials in batteries, and hence lots of poten-
tial combinations to explore! Find the optimal combination, and you have
an EV battery that can charge faster and be more efficient; but finding that
optimal combination can take years (and may involve a lot of trial and error).
Aionics is using GenAI to accelerate discovery and even create new molecules
designed for certain applications.20

184
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

The Impact of Generative AI on Digital Twins

A digital twin is a digital copy of a real-­world physical asset (such as a product,
process, or system) that can be used to run virtual simulations. The idea behind
a digital twin is to let us see what might happen if we were to make certain
adjustments in real life. The adjustments can be trialed on the digital twin with-
out having to test potentially expensive changes on the real-­world counterpart.

Digital twins have been a trending technology for a few years now, used in
industries as diverse as manufacturing (e.g., to optimize machine perfor-
mance) and Formula 1 (to test different car setups). Creating and using digi-
tal twins has, until now, been a pretty technical process. But with GenAI, in
theory, you can ask a GenAI tool to create the digital twin for you, making it
so much easier for organizations to harness digital twin technology and cre-
ate simulations of real-­world products, objects, and systems. So, for example,
you’ll be able to create a digital twin of your product to see how it performs
under many different circumstances without having to conduct expensive
and time-­consuming tests on the physical product.

Importantly, you will be able to interact with simulations and digital twins
using natural language technology. For example, you could ask it, “now test
the product in temperatures below zero,” or whatever you like. And in the case
of a digital twin of an entire factory floor – with the digital twin reflecting real-­
world performance – a supervisor could simply ask the system, “Hey, what’s
happening on the floor today?” or “What are the top three predictive main-
tenance issues that need our attention today?” The supervisor could oversee
their operations via a real-­time 3D computer representation. (The same idea
applies to all sorts of premises or operations – factories, retail premises, sports
stadiums, energy grids, you name it.) It’s yet another example of how GenAI
promises to make technology more accessible.

Aiding Academic Research

Before we close out this chapter, I wanted to briefly mention academic
research as another area where GenAI can add enormous value. In fact, new
tools based on large language models are already emerging to aid academic

185
 GENERATIVE AI IN PRACTICE

research – for example, by scanning through thousands of research articles

for information and providing summaries.

One example comes from a tool called “Consensus,” which provides answers
to yes/no questions based on the consensus of the academic community. So
you can ask a question like, “Is immigration good for the economy?” and it
will reply with the academic consensus (in this case, studies show that immi-
gration is generally an economic benefit). It will also provide a citation list
and summaries of the academic articles used in its analysis. So far, Consensus
focuses on six topics: economics, sleep, social policy, medicine, mental health,
and health supplements.

Another tool, called “Elicit,” can find relevant academic papers for you based
on a specific question you have. It will then summarize the key information
from those papers. It’s described as an “AI research assistant.” Scite does a
similar thing, unearthing relevant academic papers.

It’s easy to see how GenAI could transform academic study and research in
future, and even allow non-­academics to access academic information in a
more user-­friendly way.

Key Takeaways
In this chapter, we’ve learned:

• GenAI will shape the drug therapies of the future by accelerating the
drug discovery process and helping to bring drugs to clinical trials
much faster – potentially years faster.
• In design, generative design (which uses AI to optimize the design
process) has been around for a few years, but the addition of GenAI
interfaces promises to improve AI-­aided design even further. GenAI
can, among other things, help designers work with complex genera-
tive design tools more easily, generate endless potential designs accord-
ing to parameters and criteria, simulate products, and accelerate the
product development lifecycle. This can apply to all sorts of fields, from
­jewelry to automotive design, to building design.

186
 CRAFTING TOMORROW: AI IN DESIGN AND DEVELOPMENT

• Related to this, GenAI will also enable companies to create digital

twins more easily for simulations of real-­world products, objects, and
­systems – using natural language prompts.
• And finally, GenAI is also making academic research more accessible.
New tools are emerging to help answer questions based on the consen-
sus of the academic community, and provide easy-­to-­digest summaries
of research papers.

Clearly GenAI is going to positively disrupt the fields of design, development,

and manufacturing. But now let’s turn to another sector that’s set to be trans-
formed by GenAI –banking and finance.

Notes
1. AI’s potential to accelerate drug discovery needs a reality check; Nature; https://
www.nature.com/articles/d41586-­023-­03172-­6
2. How Generative AI Is Revolutionizing Healthcare & Drug Discovery; Bernard
Marr, YouTube; https://www.youtube.com/watch?v=dIOBS-­T0Z-­g
3. Gen AI for the Genome: LLM Predicts Characteristics of COVID Variants;
NVIDIA; https://blogs.nvidia.com/blog/generative-­ai-­covid-­genome-­sequences/
4. Recursion Bridges the Protein and Chemical Space with Massive Protein-­
Ligand Interaction Predictions Spanning 36 Billion Compounds; Recursion;
https://ir.recursion.com/news-­releases/news-­release-­details/recursion-­bridges-­
protein-­and-­chemical-­space-­massive-­protein
5. Quicker Cures: How Insilico Medicine Uses Generative AI to Accelerate Drug
Discovery; NVIDIA; https://blogs.nvidia.com/blog/insilico-­medicine-­uses-­
generative-­ai-­to-­accelerate-­drug-­discovery/
6. New AI-­generated COVID drug enters Phase I clinical trials: ‘Effective against
all variants’; Fox News; https://www.foxnews.com/health/new-­ai-­generated-­
covid-­drug-­enters-­phase-­i-­clinical-­trials-­effective-­variants
7. Aiddison: Harnessing generative ai to revolutionize drug discoveries; Merck;
https://www.merckgroup.com/en/research/science-­s pace/envisioning-­
tomorrow/future-­of-­scientific-­work/aiddison.html
8. Gen AI may power the next generation of immunotherapies; Omnia Health;
https://insights.omnia-­h ealth.com/technology/gen-­a i-­m ay-­p ower-­n ext-­
generation-­immunotherapies
9. Dell partners with the University of Limerick to develop an AI platform for can-
cer research; Windows Central; https://www.windowscentral.com/hardware/
dell/dell-­university-­of-­limerick-­ai-­for-­cancer-­research

187
 GENERATIVE AI IN PRACTICE

10. Google DeepMind Introduces a New AI Tool that Classifies the Effects of
71 Million ‘Missense” Mutations; Market Tech Post; https://www.marktechpost
.com/2023/09/23/google-­deepmind-­introduces-­a-­new-­ai-­tool-­that-­classifies-­
the-­effects-­of-­71-­million-­missense-­mutations/
11. AI suggested 40,000 new possible chemical weapons in just six hours; The Verge;
https://www.theverge.com/2022/3/17/22983197/ai-­new-­p ossible-­chemical-­
weapons-­generative-­models-­vx
12. 4 examples of generative design in action; Archistar; https://www.archistar.ai/
blog/4-­examples-­of-­generative-­design-­in-­action/
13. Autodesk’s Generative Design: Optimizing Design Through AI; Direct Indus-
try; https://emag.directindustry.com/2023/03/15/autodesks-­generative-­design-­
optimizing-­design-­through-­ai/
14. Microsoft and Siemens Team Up to Introduce a GenAI Assistant for Manufac-
turing; Investopedia; https://www.investopedia.com/microsoft-­and-­siemens-­
team-­up-­to-­introduce-­a-­genai-­assistant-­for-­manufacturing-­8384823
15. Technology Could Turn You into a Tiffany; NY Times; https://www.nytimes
.com/2021/04/23/fashion/jewelry-­technology-­augmented-­reality.html
16. Technology Could Turn You into a Tiffany; NY Times; https://www.nytimes
.com/2021/04/23/fashion/jewelry-­technology-­augmented-­reality.html
17. How Nike used algorithms to help design its latest running shoe; Wired; https://
www.wired.co.uk/article/nike-­epic-­react-­flyknit-­price-­new-­shoe
18. Fashable reimagines the future of fashion design with Azure Machine
Learning and Pytorch; Microsoft; https://customers.microsoft.com/en-­us/
story/1558909662014453187-­fashable-­retail-­azure
19. Toyota’s Research Institute develops new AI technique with potential to help
speed up vehicle design; Toyota; https://media.toyota.co.uk/toyota-­research-­
institute-­d evelops-­n ew-­a i-­t echnique-­w ith-­p otential-­t o-­h elp-­s peed-­u p-­
vehicle-­design/
20. How generative AI is creeping into EV battery development; TechCrunch; https://
techcrunch.com/2023/10/14/how-­generative-­ai-­is-­creeping-­into-­ev-­battery-­
development

188
 15
BANKING AND FINANCIAL
SERVICES: AI AS A
DISRUPTIVE FORCE

AI has always been a game-­changer in finance, but GenAI will take this to a
whole new level. GenAI’s ability to create content, analyze data, simulate sce-
narios, and optimize processes will enable financial institutions to make more
informed decisions, innovate in product design, improve the services deliv-
ered to customers, enhance risk management practices, and maintain compli-
ance more efficiently. As such, GenAI will profoundly change the landscape
of banking operations.

As we’ll see in this chapter, banks and other financial institutions are already
starting to implement GenAI. It’s not something they’re shying away from.
In fact, I think GenAI will end up bringing one of the biggest shakeups the
financial world has seen – for example, by democratizing financial advice for
everyone. According to McKinsey, GenAI could add between US$200 and
US$340 billion in value for banks per year (9–15% of operating profits).1 Safe
to say, any bank that doesn’t embrace GenAI, risks being left behind.

Generative AI Uses in Banking and

Financial Services
So how exactly can banks and financial services companies use GenAI? Let’s
explore the main use cases from this industry.

189
 GENERATIVE AI IN PRACTICE

General use cases and possibilities

We’ll get to real-­world examples later in the chapter, but for now let’s explore
some overarching uses for GenAI.

One of the obvious uses is in customer services – such as banking chatbots

that can answer questions and respond to complaints 24/7. GenAI can also
help to onboard new customers in a more efficient way, for example, by using
a tool like ChatGPT to help customers fill out forms correctly and answer
their questions about switching banks. We talked about customer service a lot
in Chapter 8, so do circle back there if you want a refresh, but the key takea-
way in relation to banking is this: GenAI will give rise to a new era of conver-
sational banking that simulates human conversations. More like the way we
used to do banking.

Just as with other sectors, personalization is a major use case in banking,

such as personalized tools and advice, designed to help customers use their
money more wisely. GenAI can deliver in-­depth insights into customer
behavior and preferences, and this will help banks create new products and
services that are more thoughtful and customized. By simulating how dif-
ferent product and service features meet the needs of individual customers
or market segments, GenAI can help banks create highly tailored products
and additional services that better serve their customers. A good example
comes from Square financial services company, which we’ll look at later in
the chapter (Hint: they’re using GenAI to help business customers generate
content for their business).

Another major use case is automating many of the behind-­the-­scenes pro-

cesses and operations in financial institutions, such as conducting financial
analysis and writing reports. This may include:

• Credit decisioning: GenAI can produce more nuanced credit ­models

that consider a plethora of variables, including unconventional data
sources that might be predictive of creditworthiness. By generating
complex behavioral profiles, these models could potentially reveal
insights that lead to safer (and more equitable) lending.

190
 BANKING AND FINANCIAL SERVICES: AI AS A DISRUPTIVE FORCE

• Denial explanations: As well as aiding the credit decision process,

GenAI can be used to generate user-­friendly explanations of why a cer-
tain decision was made. Not only does this help customers with future
applications; it builds trust in AI decision-­making.
• Fraud detection frameworks: GenAI can be employed to simulate pat-
terns of fraudulent behavior, which can train fraud detection systems to
recognize and block sophisticated scams and security breaches before
they happen. And later in the chapter, we’ll see how GenAI is help-
ing banks and payment providers search communications for poten-
tial scams.
• Payment processing systems: As well as helping to detect fraud, GenAI
can optimize payment processing systems in other ways, for example,
optimizing cross-­border payments.
• Financial modeling and scenario analysis: GenAI can create sophisti-
cated financial models that simulate various economic and market con-
ditions to predict their effects on investment portfolios, asset values, or
institutional risk exposure. Banks can use these models to prepare for
a range of future scenarios, making their operations more resilient to
shocks and stresses.
• Quantitative trading: GenAI algorithms can create numerous strat-
egies and test them against historical and simulated future market
data. They can generate predictive models that improve over time,
potentially offering higher returns or lower risk on investment
strategies.
• Asset and portfolio management: GenAI could significantly enhance
asset management by simulating the performance of different asset
combinations under a range of market scenarios, thus generating opti-
mized portfolios based on predicted future performance.
• Regulatory compliance: GenAI can be instrumental in developing
compliance systems that continuously learn and adapt. It can generate
and test thousands of compliance scenarios to ensure that new financial
products or services meet stringent regulatory standards across differ-
ent jurisdictions.

191
 GENERATIVE AI IN PRACTICE

• Know Your Customer (KYC) and Anti–Money Laundering (AML):

These processes are critical for banks to mitigate financial crime and
maintain regulatory compliance. GenAI can assist in automating these
processes by analyzing large amounts of customer data, including per-
sonal data and transaction history, and identifying potential compli-
ance issues.
• Operational risk management: By simulating various operational
failure scenarios, GenAI can help in forecasting and mitigating
potential operational risks, leading to more robust risk management
practices.
• Legacy software maintenance: Did you know that many banks still
rely on extremely outdated software written in legacy programming
languages like COBOL (which is a 60-­year-­old programming lan-
guage)? Not many developers are fluent in these old languages, which
can make maintenance an issue. But GenAI can easily be trained
to understand these legacy programming languages and maintain
older software.

Supporting accountants and internal accounting functions

GenAI can also play a vital role in aiding accountancy firms and internal
accounting functions within organizations. While this won’t be the main
focus of this chapter, I wanted to highlight that GenAI can help accounting
teams augment accounting and finance processes, such as:

• Creating and summarizing reports

• Forecasting financials
• Analyzing data
• Mitigating risks
• Delivering user-­friendly recommendations for business leaders
• Generating simulations of various scenarios to aid decision-­making
• Automating everyday processes, such as generating invoices

192
 BANKING AND FINANCIAL SERVICES: AI AS A DISRUPTIVE FORCE

As with other sectors, we can expect to see more AI tools emerge to support
accounting professionals in their work – for example, by helping them create
more forward-­looking insights that drive business performance.

The Risks of Generative AI

GenAI will bring major changes to the financial sector. It will also bring new
risks. A report by the International Monetary Fund cites four main areas of
risk for the financial services industry:2

• Inherent technology risk: This includes factors like data privacy (e.g.,
leakage of personal data through training data), and the risk of embed-
ded bias in data perpetuating discrimination.

• Performance risk: This includes the risk of GenAI systems “hallu-

cinating” incorrect (but convincing) information, and the problem
of “explainability” (which is particularly important because financial
institutions must be able to explain the reasoning behind their actions
and decisions).

• Cybersecurity threats: This spans both the possibility of a GenAI sys-

tem being attacked, and the use of GenAI by hackers to target finan-
cial systems.

• Financial stability risk: The IMF suggests that GenAI could pose a
systemic risk, especially if advisors over-­rely on the technology. One
example given is GenAI systems potentially encouraging financial
institutions to follow the same decision-­making process (“herd behav-
ior”), thereby impacting market liquidity.

Despite these threats, it’s worth noting that the IMF still believes GenAI shows
great promise for the financial sector, helping to drive efficiency, increase
compliance, and enhance the customer experience. However, GenAI tools
must be deployed with caution and careful (human) oversight to realize their
full potential.

193
 GENERATIVE AI IN PRACTICE

Examples from the Financial Sector

Now let’s explore some real-­world examples of how banks and financial ser-
vice providers are harnessing GenAI:

Helping banks (and insurance companies) provide

a better service
Let’s kick off with British bank, NatWest, which has collaborated with IBM to
deliver enhancements to the bank’s virtual assistant, Cora.3 Thanks to GenAI,
Cora+ will be able to provide NatWest customers with a wider range of infor-
mation, all through personalized, conversational interactions. The idea is to
provide a more human experience for customers looking for information or
looking to compare products and services across NatWest’s offering.

Private Iranian bank, Saman Bank, is also using GenAI to provide personal-
ized financial solutions. It has developed Saman Bot, a virtual AI assistant
that can answer customer queries and deliver financial advice 24/7.4 These are
early examples of banks deploying GenAI in customer-­facing solutions, but
they certainly won’t be the last.

Square, the financial services platform owned by Block Inc., is embracing

GenAI in a really interesting way – using the technology to provide new tools
for customer content creation, onboarding, and setup.5 Among the new fea-
tures is a Menu Generator tool, which allows restaurants to create a menu in
minutes for their website. There’s also a GenAI-­powered email feature, which
helps businesses create personalized messages for customers. There’s even a
website copy generator, which can write headlines, copy, and blog posts for
business websites. This gives us a taste of how banks could use GenAI to cre-
ate time-­saving new tools and products for their business customers in future.

Another example comes from insurance and financial services company,

Nationwide, which has been using GenAI to improve pet insurance provi-
sion by helping pet owners take better care of their pets’ health. The com-
pany turned to GenAI to create easy-­ to-­
understand summaries for 71
different health conditions – information that was used by Nationwide’s Pet
Health Zone platform, which uses pet claims data to provide personalized

194
 BANKING AND FINANCIAL SERVICES: AI AS A DISRUPTIVE FORCE

information about pet health risks. Thanks to GenAI, the team accelerated
the creation of 35 000 words of medical content, saving more than 300 hours.6

Streamlining banking operations with GenAI

Meanwhile, several banks in China are reportedly embracing GenAI for behind-­
the-­scenes functions. For example, Industrial and Commercial Bank of China is
exploring GenAI language models as a wealth management tool. Fintech giant,
Ant Group, is doing something similar; developing its own language model for
wealth management and insurance that can assist finance professionals.7

In the US, KeyBank is taking a similar approach – using GenAI to streamline

internal operations before deploying it across customer-­facing channels. The
bank is reportedly using GenAI to support its documentation teams to mini-
mize their manual workload.8

Global bank, HSBC, isn’t shying away from GenAI, either; the bank has
reportedly unearthed a “few hundred” use cases for GenAI that could poten-
tially be deployed.9

Asset management company, Schroders, has built its own internal version
of ChatGPT, trained on its own proprietary data and that is proving useful
in a number of ways. One of those is in translating content into other lan-
guages. The company says translations that previously took two days can now
be done in two minutes, and the translations have, over time, become “near
perfect.”10 Schroders is also exploring how GenAI can be used to analyze data
and identify future needs that may inform product design.

GenAI tools as a “co-­pilot” for financial professionals

Morgan Stanley has been experimenting with OpenAI’s GPT-­4 model, using
the language model as a sort of encyclopedia for financial advisors. The
company trained GPT-­4 on content from over 100 000 documents, to help
financial advisors answers questions on investment recommendations, gen-
eral business performance, and internal processes (such as how to complete
certain applications).11 Several times throughout this book we’ve seen how

195
 GENERATIVE AI IN PRACTICE

GenAI can serve as a “co-­pilot” for professionals, helping them do their job in
a more efficient way – and the financial industry is no exception.

Meanwhile, Moody’s, the company best known for risk assessment and credit
ratings, is partnering with Google Cloud to help financial service profession-
als leverage large language models to accelerate financial analysis. In essence,
the idea is to use GenAI to help professionals gain new financial insights and
summarize financial data in a quicker, easier way.12 As an example, users will
be able to easily analyze and extract “decision-­ready insight” directly from
financial disclosures and other financial reports.

Enhancing the payments industry

Examples are also emerging of using GenAI to facilitate payments. Master-
card, for instance, already uses AI to secure billions of transactions annually,
and helps banks predict or spot fraudulent activity. But the payments provider
is now actively looking to use GenAI to improve the customer experience, aid
product testing, and generally help customers solve business challenges (such
as reducing bias in credit decisions).13

Elsewhere, Stripe was an early collaborator with OpenAI on its GPT-­4 large lan-
guage model. Stripe employees were encouraged to come up with exciting ways
to use the model to identify new products and improve workflows, and one of
the uses they came up with is monitoring for fraud. Stripe has now started using
GPT-­4 to analyze posts in Stripe forums on platforms like ­Discord. (These
forums can be used by bad actors posing as legitimate community members
to try and defraud people or get them to give up personal information.) Using
GenAI-­powered syntax analysis, Stripe can now monitor these communica-
tions for potentially dodgy interactions – with GPT-­4 flagging posts that require
investigation so they can be passed on to Stripe’s fraud team.14

Stripe is also building new user tools based on the GPT model.15 One of these
is Stripe Docs, a GPT-­powered tool that allows developers to pose natural
language queries to the language model, rather than having to wade through
documentation. Stripe says this allows developers to spend less time reading
and more time building solutions.

196
 BANKING AND FINANCIAL SERVICES: AI AS A DISRUPTIVE FORCE

GenAI in accounting
I’ve also mentioned that GenAI can assist accountants and internal finance
functions in their work. As such, accounting software companies are keen
to integrate GenAI into their products. One example comes from Intuit, the
software giant behind QuickBooks online accounting software, which has
introduced an AI helper called “Intuit Assist.”16 The tool can deliver person-
alized insights and recommendations to help users solve problems (routing
the customer to a human expert for issues that need in-­depth guidance). For
example, for individuals using Intuit software to manage their personal taxes,
Intuit Assist can create a personalized tax checklist to help them file their
tax return and access refunds more easily. And for small businesses using
QuickBooks accounting software, Intuit Assist can examine customer behav-
ior and business performance to pinpoint potential cash flow issues, highlight
the most successful products, and more. And this can all be done via natural
language queries, such as, “Which invoices are overdue?”

Oracle subsidiary, NetSuite, is a provider of cloud-­based enterprise resource

planning (ERP) applications. Now, NetSuite has added a GenAI feature called
“NetSuite Text Enhance” to help finance teams (and HR, supply chain and
sales team) generate contextual and personalized content.17 The tool can gen-
erate draft text for team members to review, based on just a few starter words.
It’s one of several AI-­powered features being rolled out by NetSuite, including a
new feature that helps businesses automate data analysis for preparing budgets.

Key Takeaways
In this chapter, we’ve learned that the financial sector is set for a major shakeup
thanks to GenAI:

• AI has been a game-­changer within the financial sector for many years,
but GenAI will take this to new heights. GenAI can help banks deliver
a better, more thoughtful service to customers through AI assistants.
It can also be used to develop customized products and services, and
deliver personalized advice to clients. And it can be used to automate
and augment internal processes – such as credit decisioning, detecting
fraud, asset management, and compliance.

197
 GENERATIVE AI IN PRACTICE

• GenAI can also be used by accountants and internal finance teams

to augment their work. For example, GenAI can be used to forecast
financials, generate or summarize reports, and provide more forward-­
looking recommendations.
• However, GenAI does bring new risks to the financial sector. Spe-
cifically, there are risks around data privacy, data bias, the risk of AI
systems hallucinating incorrect financial information, cybersecurity
threats, and risks to financial stability. Therefore, GenAI tools must be
deployed with careful human oversight.
• GenAI is already being used by banks and financial services providers
in a number of different ways – from customer-­facing chatbots that
give advice 24/7, to behind-­the-­scene operations, to AI “co-­pilot” tools
for financial advisors. It is also being deployed by payment providers
such as Mastercard, and being built into accounting solutions for indi-
viduals and businesses.
• Bottom line, any bank or financial services operator that ignores GenAI
risks being left in the dust.

Now let’s turn to coding and computer programming and see how GenAI can
support the work of developers (and lay people) in interesting new ways.

Notes
1. The economic potential of generative AI: The next productivity frontier;
McKinsey & Company; https://www.mckinsey.com/capabilities/mckinsey-­
digital/our-­i nsights/the-­e conomic-­p otential-­o f-­g enerative-­a i-­t he-­n ext-­
productivity-­frontier
2. Generative Artificial Intelligence in Finance: Risk Considerations; IMF; https://
www.imf.org/en/Publications/fintech-­notes/Issues/2023/08/18/Generative-­
Artificial-­Intelligence-­in-­Finance-­Risk-­Considerations-­537570
3. NatWest and IBM collaborate on generative AI initiative to enhance cus-
tomer experience; PR Newswire; https://www.prnewswire.com/news-­releases/
natwest-­and-­ibm-­collaborate-­on-­generative-­ai-­initiative-­to-­enhance-­customer-­
experience-­301977636.html
4. Revolutionising banking: The role of AI in shaping the future; Zawya; https://
www.zawya.com/en/opinion/business-­insights/revolutionising-­banking-­the-­
role-­of-­ai-­in-­shaping-­the-­future-­x5p6syun

198
 BANKING AND FINANCIAL SERVICES: AI AS A DISRUPTIVE FORCE

5. Square’s new AI features include a website and restaurant menu generator;

­TechCrunch; https://techcrunch.com/2023/10/18/squares-­new-­ai-­features-­
include-­a-­website-­and-­restaurant-­menu-­generator/
6. Nationwide’s innovation with generative AI across business lines; Columbus Busi-
ness First Journal; https://www.bizjournals.com/columbus/news/2023/10/30/
nationwide-­s-­innovation-­with-­generative-­ai.html
7. Chinese banks jump on AI bandwagon to cut costs; Asia Nikkei; https://asia.
nikkei.com/Spotlight/Caixin/Chinese-­b anks-­j ump-­o n-­A I-­b andwagon-­
to-­cut-­costs
8. KeyBank implements gen AI internally; Bank Automation News; https://
bankautomationnews.com/allposts/ai/keybank-­implements-­gen-­ai-­internally/
9. Now HSBC has a ‘few hundred’ potential use cases for generative AI: HK CEO;
CNBC; https://www.cnbc.com/video/2023/11/03/hsbc-­has-­a-­few-­hundred-­
potential-­use-­cases-­for-­generative-­ai-­hk-­ceo.html
10. Asset Managers Benefit from Generative AI; Markets Media; https://www
.marketsmedia.com/asset-­managers-­benefit-­from-­adopting-­generative-­ai/
11. How Morgan Stanley Is Training GPT to Help Financial Advisors; Forbes;
https://www.forbes.com/sites/tomdavenport/2023/03/20/how-­morgan-­stanley-­
is-­training-­gpt-­to-­help-­financial-­advisors/
12. Moody’s and Google Cloud Partner on Generative AI Applications Tailored
for Financial Services Professionals; PR Newswire; https://www.prnewswire
.com/news-­r eleases/moodys-­a nd-­g oogle-­c loud-­p artner-­o n-­g enerative-
­ai-­applications-­tailored-­for-­financial-­services-­professionals-­301964682.html
13. 9 new AI moves in the payments industry; American Banker; https://www.
americanbanker.com/payments/list/9-­new-­ai-­moves-­in-­the-­payments-­industry
14. How Stripe is using GPT-­4 to fight fraud; Free Think; https://www.freethink
.com/robots-­ai/stripe-­gpt4
15. Stripe and OpenAI collaborate to monetize OpenAI’s flagship products and
enhance Stripe with GPT-­4; Stripe; https://stripe.com/en-­bg/newsroom/news/
stripe-­and-­openai
16. Intuit Assist Brings Generative AI to Small Business and Consumer ­Portfolio;
Forbes; https://www.forbes.com/sites/patrickmoorhead/2023/09/11/intuit-­assist-
­b rings-­g enerative-­a i-­t o-­s mall-­b usiness-­a nd-­c onsumer-­p ortfolio/?sh=
7f6a414d29dd
17. NetSuite adds generative AI to its entire ERP suite; CIO; https://www.cio.com/
article/655912/netsuite-­adds-­generative-­ai-­to-­its-­entire-­erp-­suite.html

199
 16
CODING AND
PROGRAMMING: THE AI
REVOLUTION

This new wave of advanced GenAI large language models isn’t just capable of
writing text – they can write computer code. Which makes sense when you
think that computer code is just another type of language. This means GenAI
can aid the work of coders, programmers, and developers, and speed up the
software development process.

(If you’re new to this topic and wondering what the difference is between
coders, programmers, and developers, coders use programming languages
to tell a computer or software what to do, whereas programmers or devel-
opers are generally more experienced coders who can work across multiple
programming languages and oversee projects, including developing the logic
and mapping the project from start to finish. People often use the terms inter-
changeably, which is what I’ll do in this chapter, although technically speak-
ing, coding is a part of the overall programming process.)

But how good is GenAI at creating computer code? Well, when the Alphabet-­
owned DeepMind lab pitted its AlphaCode AI model in competition against
human coders, AlphaCode’s performance roughly corresponded to “a nov-
ice programmer with a few months to a year of training.”1 Not bad at all for
an AI. And given that GenAI’s capabilities are progressing so fast, we can
expect the technology to catch up to more experienced coders in the not-­so-­
distant future.

201
 GENERATIVE AI IN PRACTICE

Another interesting aspect of this is GenAI can also help people like me –
­someone who has relatively little knowledge of programming languages – write
computer code for various applications. Let’s explore the role of GenAI in coding.

The Impact of Generative AI on Programming

We’ll kick off this section by exploring exactly what GenAI is capable of
and then see how this may impact the work of programmers (and every-
day people).

What can GenAI do?

Developing new software is like building a house – it always seems to take
longer and cost more than you expect! GenAI can help to accelerate the pro-
cess in several ways:

• Gathering requirements: All software development starts by identify-

ing requirements (such as “User wants to do X, Y, and Z”). GenAI can
aid this process by generating a list of requirements. It can also be used
to review requirements and check nothing has been missed (e.g., secu-
rity requirements).
• Generating code: GenAI can turn natural language instructions into
functioning code. Which essentially turns English (or whatever your
native tongue is) into a computer programming language. . .
• Completing code: As coders type, the GenAI can suggest code
­completions – which saves coders time, especially when working on
repetitive or mundane tasks. Code completion can also help to reduce
human error.

GenAI can also be a useful tool when it comes to reviewing and testing
software, by:

• Reviewing code: GenAI can check through existing code and suggest
improvements or create alternatives that are more efficient. It can also

202
 CODING AND PROGRAMMING: THE AI REVOLUTION

analyze code according to coding style guidelines, to ensure code is

consistent.
• Fixing bugs: GenAI can be used to identify and fix bugs in code, to
create a better end product.
• Testing software: GenAI can carry out many of the testing phases, such
as generating test cases, generating test code, and analyzing test results.

Looking ahead, GenAI may also be able to predict how systems and software
might fail before the code goes into production – and, of course, tell develop-
ers how to fix it. That’s the goal of software company Dynatrace, which has
been building “predictive AI” to predict how systems will fail for years. In an
interview with ZDNET, co-­founder and chief technology officer of Dynatrace
Bernd Greifeneder says, the next stage is building GenAI into that process –
so the GenAI can tell coders how their code may cause faults and how to fix
it.2 “The typical request from a CIO is, please fix my system before it actually
fails,” he said. In other words, the holy grail of programming is to prevent
coding that causes faults, as opposed to having to fix faults further down the
line. It’s still early days for using GenAI in this way, but Dynatrace is certainly
working toward that holy grail.

Of course, there are issues with using GenAI in programming – particularly

when it comes to hallucinations, data security, and IP. (For example, you don’t
want your precious code hoovered up for training purposes, potentially to be
shared with other users outside of the company.) As in any industry, human
oversight and clear guardrails are needed.

So does this mean we won’t need software developers

in the future?
No. But it does mean they will increasingly work alongside AI, just as other
professionals like journalists and doctors will. While it’s true that GenAI can
be used to automate coding (thereby putting the work of junior coders poten-
tially at risk), I see an even bigger application being the use of GenAI as a
co-­pilot or co-­coder for even the most experienced programmers. Especially
when you consider that GenAI can be used to gather requirements and debug

203
 GENERATIVE AI IN PRACTICE

codes, or make sure that code adheres to certain requirements, as well as write
the code itself.

And let’s not forget that, for someone like a higher-­level developer or pro-
grammer, writing code isn’t the main thing they do all day. In fact, writing
code may take up as little as 20% of their time,3 with the rest of their time
spent on gathering requirements, testing, meetings, collaborating with users,
overseeing projects, etc.

What’s more, even when using GenAI tools to automate or help with pro-
gramming work, we’ll still need experienced developers to write detailed
prompts that get the best out of the AI. (You could argue writing detailed
prompts that tell the AI exactly what you want it to do is just another form of
programming. . .)

So rather than replacing programmers, GenAI will support the work of

programmers by automating repetitive tasks, accelerating software devel-
opment, and enhancing productivity. GenAI could thereby prove to be a
game-­changing tool for programmers. In one example, GenAI was shown
to increase developer productivity by 10%; in another, it cut the menial
work that developers are expected to do (such as internal communica-
tion and documentation) by a huge 70–80%.4 Perhaps it’s no wonder that
many developers are on board with the idea of using GenAI. According
to a survey by GitHub, 92% of US-­based developers are already using AI
coding tools, and 70% believe AI coding tools will give them an advantage
at work – with better code quality and faster completion time among the
anticipated benefits.5

GenAI will change how programming is done, in other words – and hopefully
make the work better for programmers.

This has advantages for everyday people, too

Clearly, GenAI is going to be a revolutionary tool for programmers. But it
may also prove a useful tool for everyday folks who want to, say, build an app
or create a piece of software, but don’t know any programming languages.

204
 CODING AND PROGRAMMING: THE AI REVOLUTION

With a GenAI tool, you can tell the GenAI what you want to achieve and
voila! It will deliver (in theory, at least, since we know that GenAIs can some-
times confidently give incorrect but plausible-­looking answers).

This means anyone can potentially become a coder, regardless of exper-

tise. Even me.

For my YouTube Short videos, I have to complete these annoying templates

that are super-­restrictive (e.g., requiring a specific number of characters for
the title, description, and so on). Doing this for each video (I create a lot of
content) and sticking to the character restriction was proving really tricky and
time consuming. So I asked ChatGPT to write me a little program to help me
create HTML templates for my YouTube videos. And it works. And it saves
me precious time. And I haven’t got any real programming experience. . .

Adoption and New Coding Tools:

Real-­World Examples
Now that we know what GenAI is capable of, let’s see how companies are
putting this into practice – starting with companies who have deployed
GenAI for their software development, and then looking at some of the
new tools that promise to help people (professionals and non-­professionals)
create code.

Examples of companies using GenAI internally

Coders at software company, Freshworks, have been using ChatGPT to write
code – and, in the process, cut the development time from around 10 weeks
down to less than a week.6 ChatGPT can write code in a number of program-
ming languages, including JavaScript, Python, and C++. It can also act like a
coding tutor, explaining how code works, and debug code.

Collaboration software company, Augmend, is using Microsoft’s GitHub

Copilot tool and ChatGPT for development, debugging, and learning –
­helping the small start-­up accelerate the software development cycle and be

205
 GENERATIVE AI IN PRACTICE

more competitive. As CEO and co-­founder, Diamond Bishop, puts it, “Going
back to coding without this augmentation feels a lot like working without
the internet at your fingertips.”7 He also noted that GenAI tools have nearly
doubled the team’s productivity – which is impressive for a team of just five
developers.

Meanwhile, at cloud technology company, Oracle, CTO and founder, Larry

Ellison, has said Oracle will develop future applications using GenAI, based
on developer prompts. “We’re not writing it anymore. We’re generating that
code. It fundamentally changes how we build applications, how we run
applications. It just changes everything.”8 He also noted that GenAI enables
faster development with smaller development teams, and reduced secu-
rity flaws.

GenAI is also proving useful for non-­technology businesses (if there’s any
such thing as a non-­technology business these days). Dubai Electricity and
Water Authority (DEWA) has adopted the GitHub Copilot tool, using it to
help its software developers write code and develop applications.9 As the
name suggests, Copilot acts like a co-­pilot, giving suggestions and guidance
to programmers writing code to help speed up the process.

Seattle-­based real estate company, Redfin, has used large language models
(LLMs) including ChatGPT to help with programming tasks. As Redfin
CTO, Bridget Frey, says, GenAI has proven useful for “migrating from one
programming language to another, helping developers understand legacy
code written by other colleagues, or writing functions for converting data
formats. These are good examples of tasks that our engineers can do without
the assistance of LLMs, but with these models, they can move much faster.
Something that used to take an engineer 30 minutes to do can now be done
by AI in one minute.”10

Meanwhile, at General Motors, Chief Data and Analytics Officer, Jon ­Francis,
has said AI is a huge area of investment for the auto giant, noting that it
will be used “in the back office with HR chatbots, on the factory floor with
predictive maintenance, and in IT operations by scaling and productionizing
software development.”11

206
 CODING AND PROGRAMMING: THE AI REVOLUTION

New coding tools are emerging

Now let’s take a quick look at some of the many tools that are emerging to help
coders code. Obviously ChatGPT is one that you’ve already heard of, but let’s
explore a wider list (in alphabetical order, not order of preference):

• Auto-­GPT: This open-­source tool harnesses OpenAI’s GPT-­4 model to

perform tasks – including, for example, creating a simple app. Ask it to
create an app and not only will it break the project down into specific
tasks, it will autonomously complete the tasks.
• Bard: Google’s ChatGPT, basically. Bard can function as a coding assis-
tant, generating code from prompts, and explaining code pasted in.
It’s not explicitly designed for this (see Google Codey below), but it
can do it.
• ChatGPT: OpenAI’s ChatGPT wasn’t designed to generate code, but
it can write code in various languages when prompted. Like Bard, it’s
perhaps not the best tool out there for programming, but the easy-­to-­
use interface makes it a decent starting point for those experimenting
with AI-­generated code.
• Codacy: An AI code quality/code analysis tool that helps to auto-
mate code analysis. It’s designed to help developers create better soft-
ware, faster.
• Code Llama: Meta (Facebook) fine-­tuned its Llama 2 LLM to create
this tool for coding in various programming languages.
• CodeWhisperer: This is Amazon’s code generation tool, developed
using open-­source code and Amazon’s own data. The tool takes natural
language prompts and creates code based on the programmer’s objec-
tives, in their style. It can also be used for code completion.
• CodeWP: WordPress’s code generation tool is designed for develop-
ers who build sites on WordPress’s hugely popular content manage-
ment system.
• Codey: Google’s AI coding assistant is designed to streamline work-
flows and increase productivity – as such, it can do the things you’d

207
 GENERATIVE AI IN PRACTICE

expect, including code generation and code completion. Its “code chat”
function lets developers chat with a bot to get help with learning new
concepts, answering questions about code, and debugging.
• GitHub Copilot: Microsoft’s GenAI for writing code is a code-­
suggestion and generation tool, designed to enhance the work of devel-
opers. One of the ways it can do this is by suggesting code snippets as
developers’ type. It was created in collaboration with OpenAI.
• SAP Build Code: SAP’s tool for application development is optimized
for Java and JavaScript, and can be used for coding, testing, integra-
tions, and application lifecycle management.
• Salesforce Anypoint Code Builder: This code builder tool is designed
to help developers reduce costs and speed up software development. So
far, so standard. But Salesforce also offers industry-­specific integration
via packaged tools for sectors like healthcare and financial services.
• Tabnine: A coding assistant built on OpenAI’s Codex model, Tabnine,
can generate code suggestions, autocomplete lines of code, and match
in-­house conventions.
• Tynker Copilot: Designed for coders aged 6–12, Tynker Copilot applies
GenAI to coding education. This game-­based coding platform, which
was created by fine-­tuning Meta’s Llama 2 LLM, helps young coders
turn their imaginative ideas into visual block code for apps and games.
• Uizard: This tool uses GenAI to help people design mobile apps, web-
sites, and landing pages in minutes. With its intuitive interface, it sim-
plifies the design process enabling quick and easy prototyping.
• Watsonx Code Assistant: IBM’s GenAI tool is designed to speed up
code generation and increase developer productivity.
• What The Diff: A GenAI-­powered code review assistant, What The
Diff reviews the “diff ” of pull requests. A pull request is a way for devel-
opers to notify colleagues that they have completed something (such
as a bug fix) and are ready to begin merging new code changes into
a project. This tool explains the differences between the new and old
code in plain English, so everyone in the team (including non-­techies)
can understand what’s been done.

208
 CODING AND PROGRAMMING: THE AI REVOLUTION

This is by no means an exhaustive list of the many coding assistant tools com-
ing onto the market. I just wanted to give you a flavor of what’s available, from
huge tech corporations and lesser-­known companies alike. As GenAI tools
like these become more commonplace, developers will increasingly incorpo-
rate them into their everyday work – hopefully making the work of develop-
ers easier and better.

Key Takeaways
In this chapter, we’ve learned that GenAI can serve as a useful tool in software
development:

• GenAI can automate or assist in the writing, reviewing, and testing of

computer code.
• Rather than making developers extinct, GenAI will support the work of
developers by automating repetitive tasks, accelerating software devel-
opment, and enhancing productivity. Examples have shown how GenAI
can accelerate the development process from weeks down to days.
• Tech and non-­tech companies are beginning to use GenAI for their
software development tasks, and new GenAI coding assistant tools are
rapidly emerging to help developers create better code, faster.

Now let’s turn to another somewhat unexpected skill of GenAI – analyzing

data. Turn to the next chapter to see how GenAI can help us solve some of the
biggest problems around working with (and understanding) data.

Notes
1. Can coding academies survive the AI era?; Fast Company; https://www.fast
company.com/90931184/can-­coding-­academies-­survive-­the-­ai-­era
2. AI aims to predict and fix developer coding errors before disaster strikes; ZDNet;
https://www.zdnet.com/article/ai-­aims-­to-­predict-­and-­fix-­developer-­coding-­
errors-­before-­disaster-­strikes/
3. AI Won’t Eliminate Programming, But Can Make It Better; Inside Big Data;
https://insidebigdata.com/2023/11/13/ai-­wont-­eliminate-­programming-­but-­
can-­make-­it-­better/

209
 GENERATIVE AI IN PRACTICE

4. How Generative AI Can Increase Developer Productivity Now; The New

Stack; https://thenewstack.io/how-­g enerative-­a i-­c an-­i ncrease-­d eveloper-­
productivity-­now/
5. Survey reveals AI’s impact on the developer experience; GitHub; https://github.
blog/2023-­06-­13-­survey-­reveals-­ais-­impact-­on-­the-­developer-­experience/
6. Software company CEO says using ChatGPT cuts the time it takes to complete
coding tasks from around 9 weeks to just a few days; Business Insider; https://
www.businessinsider.com/chatgpt-­c oding-­openai-­c eo-­s ave-­t ime-­a i-­j obs-­
software-­2023-­5?r=US&IR=T
7. AI and coding: How these tech companies are using generative AI for program-
ming; Geek Wire; https://www.geekwire.com/2023/ai-­and-­coding-­how-­seattle-­
tech-­companies-­are-­using-­generative-­ai-­for-­programming/
8. Larry Ellison outlines Oracle’s generative AI strategy; Oracle; https://www
.oracle.com/artificial-­intelligence/larry-­ellison-­cloudworld-­genai-­strategy/
9. DEWA adopts new Microsoft generative AI tool to drive digital transformation;
Zawya; https://www.zawya.com/en/business/technology-­and-­telecom/dewa-
­adopts-­new-­microsoft-­generative-­ai-­tool-­to-­d rive-­d igital-­t ransformation-­
yh60rr4j
10. AI and coding: How these tech companies are using generative AI for program-
ming; Geek Wire; https://www.geekwire.com/2023/ai-­and-­coding-­how-­seattle-­
tech-­companies-­are-­using-­generative-­ai-­for-­programming/
11. Every enterprise plans to increase AI spending next year; ZDNet; https://www
.zdnet.com/article/every-­enterprise-­plans-­to-­increase-­ai-­spending-­next-­year/

210
 17
DATA INSIGHTS:
HARNESSING THE POWER
OF GENERATIVE AI

Data is a vital asset for today’s businesses. It is the fuel that powers better
decision-­making in organizations. As such, it’s crucial that people across the
organization are able to extract insights from data. However, that is more
­easily said than done. Rather than being empowered by data, many people
find themselves intimidated (or even paralyzed) by it.

In this chapter, we’ll explore the problem of being rich in data but poor in
insights, and discover how GenAI can help businesses get more out of data.

What Is the Data Problem and How Does

Generative AI Help?
We live in a world that’s full of data – indeed, everything we do generates data,
and the average business is absolutely swimming in the stuff. But what is the
problem with having such a vast sea of data? You can end up drowning in it.

The thorny issue of data overwhelm

Useful as data is, the sheer volume of data can simply overwhelm people.
This phenomenon is described by software leader Oracle as the “Deci-
sion Dilemma.” You could also call it “decision paralysis” or “data anxiety.”

211
 GENERATIVE AI IN PRACTICE

Whatever you call it, the basic gist is that more data causes anxiety and lack of
action, instead of better decisions.

For its 2023 Decision Dilemma report, Oracle surveyed more than 14 000
employees and business leaders across 17 countries, and the results were
eye-­opening:1

• 83% agreed that access to data is essential for helping businesses make
decisions, BUT. . .
• 86% said that data makes them feel less confident, and
• 72% said that data has stopped them being able to make a decision.

When I spoke to Oracle’s James Richardson, VP for Product Strategy, Analyt-

ics, he told me, “It’s the anxiety that comes from having the wealth of data we
have at our fingertips, but perhaps a limited amount of time to use it . . . one
of the interesting findings was a significant number of people said that at the
point of making a decision, they were so overwhelmed by the data that they
didn’t make the decision. So the question is, how do we deal with that?”

I’ve certainly noticed this myself in my own consultancy work – organizations

are rich in data but that doesn’t necessarily mean they have the insights they
need to make decisions and take action. It’s perhaps no wonder that 70% of
business leaders would sometimes prefer to simply leave these decisions in
the hands of a robot or AI.2 Richardson doesn’t believe this means that people
are looking to abdicate all responsibility for decision-­making; we’re just des-
perate for assistance. “What they are really saying is ‘help me’ ”, he says.

So what do we do about this? As we’ll see in this chapter, part of the solution
may lie in GenAI’s ability to make sense of data and extract useful informa-
tion that helps us make better decisions.

The importance of data literacy

To truly harness GenAI, we need businesses to embrace a new era of “self-­
service” analytics, where anyone has the power to find the answers they need.

212
 DATA INSIGHTS: HARNESSING THE POWER OF GENERATIVE AI

(As opposed to the traditional approach where insights would be handed

down from business analysts within the organization.) To successfully make
this transition, organizations must put resources into supporting workforces
through these changes – for example, by setting up data and analytics hubs
where those who have professional data skills can assist those who don’t.

Because, if we’re honest about the situation, there’s a startling lack of data lit-
eracy out there. (By data literacy, I don’t mean being a data analyst; I mean the
ability to confidently work with data to get the information you need.) Partly,
this lack of data literacy is down to anxiety around working with data, and the
overwhelm that can come with having so much data at your fingertips, but
it’s also down to a lack of training in data literacy. According to one telling
survey, 82% of leaders expect all employees to have basic data literacy, and
79% say their departments are equipping workers with critical data skills – yet
only 40% of employees say they’re being provided with the data skills their
employers expect.3

We urgently need people to become more confident and competent at work-

ing with data. Which is why all organizations must invest in a data literacy
program that not only provides data training, but also instills a data-­driven
culture from top to bottom. Importantly, GenAI can also play a role in increas-
ing data literacy. How? By giving anyone the ability to analyze vast amounts of
data in an intuitive way, using natural language questions.

GenAI and its ability to understand data

GenAIs are able to create outputs (such as text or images) by learning from
vast amounts of data and recognizing patterns in that data. Which is a mas-
sive simplification, of course, but the truth is even the best AI experts aren’t
able to explain exactly how GenAIs do what they do. The point is, GenAI
systems are adept at analyzing huge amounts of data, and organizations can
use this to their advantage.

GenAI has the potential to democratize data analytics by making advanced

data interpretation and insight generation accessible to companies of all sizes.
GenAI streamlines the analysis and interpretation of data by automating

213
 GENERATIVE AI IN PRACTICE

complex processes such as pattern recognition and trend analysis, allowing

for real-­time insights that turn vast datasets into easy-­to-­understand narra-
tives and visualizations. (Because with multi-­model systems that are able to
produce, say, graphs as well as text, you could have your data presented in the
exact way you want.) Plus, GenAI can in theory work with all sorts of messy,
unstructured data, including photos and video data, or social media posts –
meaning, it isn’t just limited to neatly structured data in databases.

All this makes data much more actionable for decision makers across the
organization – regardless of their data expertise. Decision paralysis, begone!
Which is good news, considering that three-­quarters of business leaders say
the daily volume of decisions they need to make has increased tenfold over
the past three years.4

As well as alleviating the problem of decision paralysis, this democratization

of data through GenAI can also help to level the playing field between large
corporations and smaller enterprises. With GenAI, you don’t need an army of
data scientists to gain competitive advantage; you just need to ask GenAI the
right questions. As such, analytics software vendors such as Microsoft, Qlik,
and Tableau are increasingly embedding GenAI in their analysis tools. But
we’ll talk more about tools later in the chapter.

Use cases for GenAI

So how exactly can GenAI be used to work with data? Potential use
cases include:

• Driving faster and better decision-­making through better insights:

Through real-­time tracking of data, decision makers can gain a better
grasp of what’s happening across the business and be presented with
actionable insights suggested by the GenAI. And this can be achieved
through natural language prompts, such as “What are our top three
customer behavior trends for the previous month?”
• Acting as a decision-­making co-­pilot: Thanks to GenAI’s conversa-
tional abilities, these tools can function as virtual advisors – a sounding
board to help discuss and generate ideas.

214
 DATA INSIGHTS: HARNESSING THE POWER OF GENERATIVE AI

• Generating summaries of data: GenAI can sift through vast quantities

of data and create executive summaries that pull out the key points,
along with best-­practice recommendations. GenAI can even be used to
suggest what information to include in reports for different audiences
(execs, department heads, managers, and so on), so that everyone gets
the information they need to make better decisions.
• Visualizing data: GenAI can generate analytics reports in an easy-­to-­
digest format –presenting insights from the data not just as text narra-
tives but also in visual format (graphs, charts, etc.).
• Automating data analytics: GenAI can potentially automate the data
analysis process and provide automatic notifications for, well, anything
you want. Spikes in sales, trending website activity, a drop in factory
machine performance, increased sick leave, you name it. . .
• Harnessing predictive capabilities: Throughout this book we’ve seen
how GenAI can help people take action before an issue becomes a real
problem. So as well as understanding what’s going on in the business
right now, GenAI can help decision makers pre-­empt what might be
coming down the line.
• Using synthetic data to test ideas and scenarios: By creating large
amounts of synthetic data that mimic real-­world data, leaders can
model scenarios that may be difficult to model with real-­world data
(e.g., because an event is a rare, but impactful, occurrence, or because
gathering that much data would be difficult and expensive). Remem-
ber that synthetic data can often be generated with just a small amount
of real-­world data as a starting point – which makes large-­scale data
analysis possible for even smaller businesses.
• Preparing data: GenAI can also be used to take care of data preparation
tasks such as tagging, classification, segmentation, and anonymization.
• Helping to clean up data for better analysis results: Because GenAI
is so good at spotting patterns, it can be used to detect anomalies and
inconsistencies in your data – things that could potentially skew results.

Again, later in the chapter we’ll look at some of the analytics tools that are
building such GenAI capabilities into their products.

215
 GENERATIVE AI IN PRACTICE

Will GenAI put data analysts out of work?

If a business leader can simply ask a GenAI analysis tool, “What do I need to
do to improve customer satisfaction?” will we really need analytics profes-
sionals in future?

I believe the answer is yes. If anything, as GenAI becomes more accessible and
mainstream, data teams may even become more critical to organizations than
they already are. After all, today’s GenAI tools still lack vital abilities like criti-
cal thinking, strategic planning, and complex problem-­solving. We will still
need people who are experts in data and who can help organizations apply
their data in the most strategic way.

Having said that, as with many jobs, the work of data analysts, business
analysts, and data scientists will undoubtedly change. Routine and repet-
itive analytics tasks will likely be automated, leaving analysts to focus on
more strategic tasks and collaborating with teams within the business. Plus,
knowledge of how to work with tools like ChatGPT will become increasingly
important for analysts.

What about the limitations of GenAI?

If this all sounds great so far, hold your horses, because, as always, there are
limitations and challenges around deploying GenAI for data analysis.

Once again, data security and privacy is a major concern here. You don’t really
want your precious organizational data to be used for training large language
models – potentially exposing information to users outside of the company.
And you’ll certainly need to take care when it comes to using GenAI for per-
sonal data, of either customers or employees. There’s also the perennial issue
of biased data, potentially skewing your results. Plus, there’s the old “black
box” issue (where we don’t really understand how these systems work and
how they arrive at answers), not to mention the risk of “hallucinations.”

That’s not to put you off using GenAI to help boost data literacy and democ-
ratize data in your organization. I absolutely think businesses should be

216
 DATA INSIGHTS: HARNESSING THE POWER OF GENERATIVE AI

exploring GenAI in this context. But it’s important to balance the technology
with human judgment and oversight. It’s also important to remember that
we’re not looking to abdicate all responsibility to machines. Which means
we still need leaders and decision makers to cultivate skills like judgment,
complex decision-­making, and strategic thinking. Read more about success-
ful implementation in Chapter 18.

Real-­World Examples and New Analytics Tools

Now let’s see how organizations have already begun to use GenAI to get more
out of their data. After that, we’ll explore some of the many analytics tools that
are now incorporating GenAI features.

Organizations using GenAI for data analysis

It’s early days for organizations adopting GenAI in their internal analytics, but
there are a couple of interesting examples we can learn from. JetBlue is on a
mission to be the “most data-­driven airline in the world,”5 with data inform-
ing every part of the business, including operations, commercial, and support
functions. To help JetBlue staff leverage data and enhance decision-­making,
the airline created an internal system called “BlueSky” that functions as an
AI-­driven operating system. JetBlue has also created an LLM called ­“BlueBot,”
which is integrated with BlueSky, that can be used by all teams at JetBlue to
access data and insights.

Elsewhere, communications company, TELUS, partnered with AI specialists,

HEAVY.AI, to improve performance analytics and the customer experience.
TELUS used interactive GenAI visualization features to extract key insights
from customer information that pinpointed upsell opportunities. As a result,
TELUS was able to focus its wireless services in the most profitable locations –
just by querying and visualizing the data in a more intelligent way.6

We’ll certainly see more companies publicizing their own GenAI success
­stories in the near future.

217
 GENERATIVE AI IN PRACTICE

GenAI-­powered analytics tools

Providers of analytics software and platforms have been building AI into their
products for years to enable more intelligent data analytics. But now they’re
increasingly adding GenAI capabilities that allow for natural language query-
ing and easy summaries of data. Let’s look at a few examples.

One example comes from Microsoft Power BI, which now incorporates
Microsoft Copilot’s large language model technology. Basically, Copilot brings
GenAI capabilities to Power BI users. This means a user can ask a question
or describe the insights or visualization they’re looking for – all in conversa-
tional language – then Copilot analyzes the data and pulls the relevant infor-
mation into a report with actionable insights.7 Users can tailor reports to their
liking, including the tone, scope, and style of narratives.

Another example comes from Terradata, which has recently launched a

new GenAI natural language interface for its VantageCloud Lake enterprise
analytics solution.8 The new capability, called “ask.ai,” allows anyone with
approved access to ask questions of their company’s data and get instant
answers. Because no complex coding or querying is needed, this expands the
use to non-­technical roles.

Qlik also offers GenAI capabilities, including fully interactive search, chat,
and multi-­language support. To expand on this, the analytics software com-
pany has launched a new suite of OpenAI connecters that essentially give
users the ChatGPT experience in the Qlik Cloud platform.9 Qlik says this
new functionality will drive broader insights and help engage more users in
data analysis.

Meanwhile, data visualization company, Tableau, is also getting on board with

GenAI, aiming to “reimagine the data experience” and bring a more “per-
sonalized, contextual, and smart” experience to users.10 Tableau AI, which is
powered by Salesforce’s Einstein AI technology, is designed to make analyzing
data as simple as asking a question – with insights delivered in plain language.
Tableau AI can even proactively suggest the questions you might want to ask
next. There’s also a new feature called “Tableau Pulse,” which will help organi-
zations make data accessible to everyone in the business, regardless of their

218
 DATA INSIGHTS: HARNESSING THE POWER OF GENERATIVE AI

level of expertise. Pulse delivers personalized insights through “data digests”

delivered inside workflow tools that people are already using, as well as a per-
sonalized metrics homepage that functions rather like a curated newsfeed.
Tableau says the homepage will become even more personalized over time as
it learns what information the user is most interested in.

Oracle has unveiled new GenAI features that allow companies to integrate
large language interfaces in their apps, enabling users to generate text from
data, summarize data, and more.11

Big data company, Alteryx, is also leveraging GenAI to help non-­skilled work-
ers better visualize and understand the trends within their data.12 With its AI
Studio, Alteryx customers can also use those insights to build new applica-
tions that streamline business operations, enhance processes, and improve
productivity – through a conversational interface.

Then there’s analytics and BI company, MicroStrategy, which has launched

MicroStrategy AI – a GenAI product that is designed to make interacting
with data faster, simpler, and more accessible for everyone.13 It includes a fea-
ture called “Auto Answers,” which allows users to engage in natural language
chats with their data, asking questions and getting instant replies. There’s also
the Auto Dashboard feature, an AI collaborator that suggests questions and
can autocomplete queries.

Kinetica, specialists in harnessing machine and sensor data, have unveiled a

feature called “SQL-­GPT,” which effectively allows users to ask anything of
their data in natural language. Type an answer, get an answer. Kinetica says
its tool outperforms other tools when it comes to accuracy and dealing with
complex questions, such as, “Show me all the aircraft over Missouri.”14

Cloud-­based data company, Snowflake, has also brought GenAI to its data
cloud platform. Snowflake Cortex allows users to chat with their data, detect
sentiment in data, extract answers, summarize information, and translate text
to a selected language.15

Elsewhere, Akkio is on a mission to bring GenAI-­powered analytics to small

and mid-­sized businesses. The company’s Generative Reports tool makes

219
 GENERATIVE AI IN PRACTICE

business intelligence accessible to SMBs by letting them quickly understand

their data, get insights, and share live reports with the team. The tool is
designed to go beyond standard dashboard tools and understand the user’s
specific problems, project requirements and use cases. In other words, it
understands the problem the user wants to solve and serves up a report to
answer questions. As Akkio describes it, “it’s like speaking with an analyst.”16

As you can tell, providers of data analytics tools are extremely keen to build
GenAI capabilities into their products. As these tools become more common-
place, users of all skill levels will be able to interrogate organizational data in
a more intuitive, easy way – and get the answers they need to do their jobs
better. Which is good news, because companies that use data to drive their
decisions will outperform those who base their decisions on instinct and gut
feeling alone.

Key Takeaways
In this chapter, we’ve learned that GenAI will help organizations solve some
of their biggest data challenges:

• By making it easier for people to interact with and query data (e.g.,
through natural language questions), GenAI promises to democratize
data, boost data literacy, and solve the problem of data anxiety.
• GenAI can automate or assist with a wide range of data analysis tasks,
including preparing data, reviewing data, analyzing data, delivering
automated notifications, creating summaries of data, surfacing impor-
tant insights, generating reports, and creating engaging visuals that
highlight insights.
• With GenAI’s help, people across the organization will be able to use
data to make better decisions and improve performance – regardless of
their technical abilities.
• However, human oversight and human skills like judgment, complex
decision-­making, and strategic thinking will still be important. The
goal is not to abdicate all decision-­making responsibility to machines!

220
 DATA INSIGHTS: HARNESSING THE POWER OF GENERATIVE AI

• Providers of analytics platforms are rapidly building GenAI features

into their products – features that allow users to ask questions of their
data, generate insights more easily, and share insights with others in
the business.

Throughout this book, we’ve touched on some of the challenges that come
with GenAI. So how can you best avoid these pitfalls and implement GenAI
in the most thoughtful way? In the next chapter, we’ll look at tips for success-
ful implementation.

Notes
1. The Decision Dilemma: How More Data Causes Anxiety and Decision Paraly-
sis; Bernard Marr; https://bernardmarr.com/the-­decision-­dilemma-­how-­more-­
data-­causes-­anxiety-­and-­decision-­paralysis/
2. The Decision Dilemma: How More Data Causes Anxiety and Decision Paraly-
sis; Bernard Marr; https://bernardmarr.com/the-­decision-­dilemma-­how-­more-­
data-­causes-­anxiety-­and-­decision-­paralysis/
3. Data literacy; Tableau; https://www.tableau.com/why-­tableau/data-­literacy
4. How AI Can Help Leaders Make Better Decisions Under Pressure; Harvard
Business Review; https://hbr.org/2023/10/how-­ai-­can-­help-­leaders-­make-­better-­
decisions-­under-­pressure
5. How JetBlue is leveraging AI, LLMs to be ‘most data-­driven airline in the world’;
Constellation Research; https://www.constellationr.com/blog-­news/insights/
how-­jetblue-­leveraging-­ai-­llms-­be-­most-­data-­driven-­airline-­world
6. 5 main uses of generative AI in business intelligence and data analytics;
­Hackernoon; https://hackernoon.com/5-­m ain-­u ses-­o f-­g enerative-­a i-­i n-­
business-­intelligence-­and-­data-­analytics
7. Introducing Microsoft Fabric and Copilot in Microsoft Power BI; Microsoft;
https://powerbi.microsoft.com/en-­us/blog/introducing-­microsoft-­fabric-­and-­
copilot-­in-­microsoft-­power-­bi/
8. Terradata Launches ask.ai, Brings Generative AI Capabilities to Vantage-
Cloud Lake; Business Wire; https://www.businesswire.com/news/home/
20230911090555/en/Teradata-­L aunches-­a sk.ai-­B rings-­G enerative-­A I-­
Capabilities-­to-­VantageCloud-­Lake
9. Experience Generative AI with Qlik; Qlik; https://www.qlik.com/blog/
experience-­generative-­ai-­with-­qlik
10. How Tableau Pulse powered by Tableau AI are reimagining the data experience;
Tableau; https://www.tableau.com/blog/tableau-­pulse-­and-­tableau-­ai

221
 GENERATIVE AI IN PRACTICE

11. Oracle Doubles Down on Generative AI Trend at CloudWorld 2023; Forbes;

https://www.forbes.com/sites/danielnewman/2023/10/03/oracle-­d oubles-­
down-­on-­generative-­ai-­trend-­at-­cloudworld-­2023/?sh=5ccf33364acf
12. Alteryx launches no-­code AI Studio to simplify the development of generative
AI applications; Silicon Angle; https://siliconangle.com/2023/10/05/alteryx-­
launches-­no-­code-­ai-­studio-­simplify-­development-­generative-­ai-­applications/
13. Introducing MicroStrategy AI: Generative AI on Trusted Data; Business Wire;
https://www.businesswire.com/news/home/20231003746515/en/Introducing-­
MicroStrategy-­AI-­Generative-­AI-­on-­Trusted-­Data
14. Ask Anything of Your Data; Kinetica; https://www.kinetica.com/blog/sqlgpt-­
ask-­anything-­of-­your-­data/
15. Snowflake’s Cortex to bring generative AI to its Data Cloud platform; ­InfoWorld;
https://www.infoworld.com/article/3709518/snowflake-­s -­c ortex-­to-­bring-­
generative-­ai-­to-­its-­data-­cloud-­platform.html
16. Akkio Launches Generative Reports to Turn Data into Decisions Instantly;
Akkio; https://www.akkio.com/press-­release/akkio-­launches-­generative-­reports

222
 Part 3
MOVING FORWARD
WITH GENERATIVE AI

Now that we’ve seen how organizations across multiple sectors are using
GenAI, it’s time for you to think about deploying the technology in your
­business – using it to enhance your products and services, boost your pro-
cesses, and more. But where do you start?

In Part 3, we’ll look at tips for implementing GenAI successfully in your busi-
ness. And in the final chapter, we’ll explore where GenAI might be headed in
the future.

223
 18
IMPLEMENTING
GENERATIVE AI: KEYS
FOR SUCCESS

Okay, you’ve seen how other companies are using GenAI to great effect. You
understand that the technology will be utterly transformative, and you’re
inspired to get started.

So now you’re ready to dive straight into the technology and start using tools
like ChatGPT for everything, right? Not quite. Because if there’s one thing I’ve
learned from my long career consulting with organizations on transformative
technologies it’s that transformation never starts with the technology itself. If
anything, technology is the last piece of the puzzle.

With GenAI, it’s not just a case of “have a play around and see how it might be
useful.” You can’t just tell everyone in your business to start using ChatGPT
right away, because there are very real challenges and limitations, especially
around privacy. You can’t have your sales team uploading customer data to a
tool like ChatGPT, for example, because that could potentially expose people’s
personal data. Yes, you want people to be using these sorts of tools as quickly
as possible, but thoughtfully. Thoughtfully, carefully, and with all the support
and guidance they need to get the best out of new technology.

If you’re going to successfully effect change – and bring people in the organi-
zation along with that change – you need the right building blocks in place

225
 GENERATIVE AI IN PRACTICE

to support that change. And the very first, arguably most important, building
block centers around culture and mindset.

Fostering the Right Culture and Mindset

for Success
Adopting such a transformative technology requires a shift in culture and
mindset. It requires an organizational culture where people in the business
continually challenge the status quo, are comfortable with change (and fail-
ure), are not afraid to experiment, and are open to learning new things.

I’m talking about a culture where people constantly ask questions like, “How
can we better serve our customers?” “How can we create more value for cus-
tomers?” “How can we create more value for the world?” and “How can we
use technology to do that?”

The GenAI mindset

I firmly believe that as this new era of GenAI unfolds, a separation will emerge
among businesses (and individuals): those who leverage the technology to
enhance innovation and productivity, and those who lag behind. The decisive
factor here is mindset.

The starting point for managing any change is usually working out how we
need to change ourselves. For most of us, that means adopting a “GenAI
mindset” (or “growth mindset” if you prefer). This goes beyond skills (which
are undoubtedly important, and we’ll talk about later in the chapter). It’s about
embracing certain beliefs and cultivating certain attributes. In particular:

• Understanding that GenAI is a tool: It won’t do our jobs for us, but we’ll use
it to do our jobs more effectively. As such, GenAI won’t replace the need for
human attributes like creativity and problem-­solving, but it will dramati-
cally cut the amount of time we spend on repetitive or mundane tasks.
• Being adaptable: This includes being willing to walk away from what
we know, even if we think it works, in order to try something new.

226
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

• Being curious: A trait that has driven every great inventor and explorer
and is vital for success in our rapidly changing workplaces. Curiosity
can be honed by training ourselves to listen actively, ask questions, and
be open when we aren’t sure about something. For me, curiosity and
humility are two sides of the same coin. If we think of curiosity as “How
can I/we do things better?”, humility says “I don’t know everything, but
am willing to learn.”
• Embracing a continuous approach to learning: Gone are the days when
school and college set us up for a lifelong career. Keeping ahead today
means constantly updating our skills and knowledge. We’ll talk more
about skills later in the chapter.
• Being willing to work collaboratively – with both humans and machines:
Because, increasingly, this is how work will get done, by combining the
best of human labor and machine labor.
• Being mindful of ethics: Most new technologies bring ethical chal-
lenges and the GenAI mindset doesn’t shy away from this. Instead, it
means continually questioning how things are done and the ethics of
those decisions.
• Thinking critically: The GenAI mindset isn’t about blindly following
what machines tell us. Far from it. More than ever, we will need humans
to think critically.

This is the same sort of mindset that’s been adopted by those who have success-
fully harnessed other big waves of transformation, like the internet. Indeed, in
his book, The Geek Way, MIT principal research scientist, Andrew McAfee,
talks about the geek mindset, describing a geek as an “obsessive maverick”
who is tenacious and unafraid to be unconventional. He sets out four charac-
teristics that define “the geek way”: speed, ownership, science, and openness.
When I spoke to him, McAfee says this mindset is what has enabled geeks
to inherit the earth (by which he’s referring to the fact that one-­third of all
market capitalization gains made by US companies in the twenty-­first century
came out of Silicon Valley).1

McAfee’s four qualities tie in nicely with what I think of as the GenAI mind-
set. Speed, for example, is about iterating quickly, failing when necessary, and

227
 GENERATIVE AI IN PRACTICE

using lessons learned to improve. Ownership means everyone is responsible

for their own work and hitting their own goals (you might also refer to this as
“autonomy”). Science means acting according to data, and making evidence-­
based decisions. And openness is the ability to accept criticism or new ideas.

All are essential elements for using GenAI successfully – and, I’d argue, all
can be aided by GenAI, as well. GenAI will help us come up with new ideas,
unearth insights in data to inform our decisions, complete our work in ways
that best suit us, and experiment.

Of course, all of these qualities aren’t just important in a work context – they’re
useful in everyday life, helping us navigate change and adversity, embrace new
experiences, get along with others, and make ethical choices. Crucially, they
are all qualities that anyone can develop and hone.

How this translates to organizational culture

(and structure)
The ideas behind mindset are closely tied to the idea of organizational
­culture – which is also often either an aid or impediment to the successful
adoption of new technologies. The GenAI mindset I’ve talked about has to
start at the top and filter down across the organization, which means you
need board awareness and buy-­in for change. You need leaders that embody
the GenAI mindset and foster a culture that embraces GenAI.

You may also need to look at your organizational structure. So many organi-
zations still have rigid, hierarchical structures. Yet, the organizations that are
most successful in this new, AI-­driven world are more porous, flexible, and
fluid. Why? Because they’re set up in a way that can roll with change and
innovate more quickly (as opposed to a hierarchical organization where deci-
sions can often take an age to get approval).

Therefore, I believe the GenAI era will accelerate the need for more fluid
and porous organizations. Where the business is organized according
to teams, rather than a hierarchical structure, and where teams may be
comprised of permanent employees in the office, employees who work

228
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

from home, and freelancers – all collaborating seamlessly to get projects

done. This means organizations will not only need to rethink traditional
approaches to employment, but also prepare workers for a new way of
working – project-­based work, with cross-­functional collaboration, using
remote and digital tools.

We also need both business leaders and individuals to accept that the concept
of a linear, lifelong career is dissolving. The future of work will be charac-
terized by fluid career paths, driven by technological advancements, societal
transformations, environmental challenges, and so on. Once again, the GenAI
mindset – one that embodies a commitment to lifelong learning, adaptabil-
ity, curiosity, etc. – will serve people well in this much more fluid working
environment.

Requiring new roles within the organization

We’ve talked about how jobs will change in the GenAI era and the emer-
gence of new roles, such as AI prompt engineer (Chapter 5). But one thing we
haven’t talked about is the requirement for new leadership roles.

In particular, I recommend organizations appoint a chief AI officer (CAIO) –

a board-­level position that promotes a better awareness of the technology
across the business (and especially the leadership team), and ensures the
technology is being used effectively. Basically, the CAIO takes oversight of
the organization’s AI strategy in order to align it with business goals. This
includes identifying opportunities to deploy AI, ensuring the skills and peo-
ple are in place to effectively execute the AI strategy, fostering an AI-­driven
culture, and dealing with issues around ethics, regulation, and compliance.

How is this different from a CDO (chief digital officer) or CTO (chief tech-
nology officer)? Much of what the CAIO does could be seen as a subset of
what a CDO or CTO does, but their time is spent specifically on tasks that
involve AI. If we consider that tasks and responsibilities involving AI are set
to become such a big part of any organization (and therefore, the CDO or
CTO’s workload), it makes sense that you need a dedicated senior-­level exec-
utive to manage it effectively.

229
 GENERATIVE AI IN PRACTICE

What if appointing a CAIO isn’t an option in your particular organization?

Some organizations are appointing AI experts as non-­executive directors to
help the board understand AI technologies and how to implement them. In
fact, this is a role I perform for a number of companies, and it’s a great option
if you don’t have (or can’t have) in-­house knowledge.

Cultivating the Right Skills and Talent

The GenAI era will prompt a complete rethink of the skills needed to succeed
in the workplace. This is yet another building block to think about when
implementing GenAI.

The art of AI delegation

In the workplaces of the future, AI delegation will become a critical skill in its
own right. What is AI delegation? It means working out what we still need to
do for ourselves, and what’s best left to machines. As GenAI becomes more
powerful and more widely accessible, how proficient we become at putting it
to work in order to drive efficiency and create value will increasingly become
a driver of success.

For business leaders and managers, this means understanding the capabili-
ties of GenAI, the tools that are available, and how they relate to your specific
business function and/or industry. And for individuals, this means looking
at your own workload and seeing how GenAI can reduce the time spent on
repetitive tasks so you have more time for other, more important tasks.

AI delegation – whether by managers or individuals – can vastly increase effi-

ciency by managing routine elements of work such as data entry, process-
ing, and analysis, detecting errors, reviewing documents, scheduling, and
time management. And as with any form of delegation, the value lies in what
can be done with the time saved – more time for tasks that require original
thought, strategic thinking, decision-­making, and relationship building.

What about other skills for success? We essentially want people to cultivate
complementary skills that help organizations get the best out of both machines

230
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

and humans. Because, remember, we’re not talking about total automation
here – we’re talking about job augmentation and making work better.

As a father of three, this is something I’m thinking about a lot lately. How will
my children succeed in this future world? What skills can we equip them with
at home to set them up for rapidly changing workplaces? It’s something we
should all be thinking about, not just for our own job, or the jobs of the people
that we lead, but also the next generation joining the workforce.

For a thorough look at essential skills, you might like to pick up a copy of my
book Future Skills: The 20 skills and competencies everyone needs to succeed
in a digital world. It takes a practical look at essential skills for our future
workplaces, with a particular emphasis on softer, human skills like empathy,
complex decision-­making, collaboration, and critical thinking – basically, the
areas where humans have the edge over machines. Once again, it’s the com-
plementary skills that will become increasingly in demand.

Tapping into AI talent (via upskilling and partnerships)

Of course, you will also need to build AI knowledge and skills across the
organization. This doesn’t necessarily mean recruiting AI talent, though. The
truth is, it will be hard for the average business to attract tech talent because
competing with the likes of OpenAI and Google for talent is nearly impossible.

Therefore, the best thing you can do is invest in upskilling your existing peo-
ple so they can understand the emerging technology, use GenAI in their work,
and decide for themselves how it can be used to add value. This approach puts
people in the driving seat, rather than make them feel like helpless passengers,
having change done to them.

Talent development is different for each company, but the most obvious
options include formal education and training programs, mentorship pro-
grams, and informal on-­the-­job learning opportunities. The good news is,
there’s never been a better time to be a lifelong learner, with a wealth of
new courses coming onto the market to educate people in AI, GenAI, and
data skills. What’s more, many of these courses are free, and you can choose

231
 GENERATIVE AI IN PRACTICE

from courses that are specific to your needs (GenAI for video ­generation,
for instance).

Online learning opportunities span the whole spectrum from massive online
learning providers like Udemy and Coursera, to courses created by tech
companies (and tech-­specific educators). For example, IBM has launched an
AI Academy, Microsoft and Amazon have created courses on GenAI, and
Google has launched both free and paid-­for GenAI training courses on its
Cloud Skills Boost platform, including courses designed for non-­technical
roles (such as sales, marketing, HR, etc.).

May I also put in a good word for my own YouTube channel, which is a great
(and free) resource for non-­ technical people grappling with technology
changes. Just search my name on YouTube to subscribe to my channel.

Developing partnerships with start-­up technology companies – and perhaps

even university departments specializing in AI – is another great way to tap
into technology expertise without hiring in-­house talent. You may also want
to consider acquiring businesses that have the in-­house skills you need. Plus,
there’s the vast pool of freelance workers out there. . .

Retaining talent will become more important than ever

Competition for talent – especially AI-­literate talent – is intense. So having
put all that thought into essential skills, and equipping people with the skills
for success, it makes sense that you really want talent to stick around. You
need to create an environment that boosts employee retention.

This encompasses all the things you’d expect, like good pay, attractive ben-
efits, flexible working arrangements, opportunities for career development,
recognition for a job well done, a welcoming work environment, and so on.
But again, it also comes down to the organizational culture and whether it
supports attributes like curiosity, continual learning, and adaptability – or
whether it squashes innovation.

In this era of rapid transformation, another good way to boost employee

retention is to hire for culture fit and potential, not just skills – because when

232
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

someone is a great fit for your organization, even if they aren’t a ­perfect
skills match, they’re more likely to be happier in the company and stay
longer. Remember, skills can be developed, but a cultural mismatch is hard
to overcome.

Data as Another Key Building Block

AI is nothing without data. As such, data must be treated as one of today’s
most valuable business assets. (And crucially, as a shared asset, with data
shared across the business. The last thing you want is to have lots of separate
pockets of data that each team hoards.)

The first point to note is that your business should already have a data strategy
in place. This is where you identify your business problems and unanswered
questions, the data you need to solve those challenges, the technology you
need to accomplish your goals, and how you will protect your data through
good governance. If you haven’t got a data strategy, I urge you to read my best-
selling book Data Strategy: How to Profit from a World of Big Data, Analytics
and the Internet of Things. And if you have got a data strategy in place, be sure
to review and update it in light of what you’ve learned about GenAI.

There is so much data available in the world today. We have more data than
ever before, and the amount of data is growing all the time. Yet, only a frac-
tion of available data is used by organizations. A thorough, up-­to-­date data
strategy will help you get the most out of this vast resource.

As well as having an up-­to-­date data strategy in place, you also need to con-
sider the shelf-­life of data. Because, like milk and reality TV stars, data can
quickly go out of date.

The importance of real-­time data

Timing is everything. It’s useful to understand what happened in your busi-
ness last week, last month, or last year. But even more important than that,
is understanding what’s happening in your business right now. Which is why
one of the biggest trends in data and analytics is real-­time insights. As you

233
 GENERATIVE AI IN PRACTICE

consider the data that will help you harness GenAI, be sure to think about
how you can capture and act on information as it happens – or as close to that
as possible.

The reason that real-­time data is so valuable to businesses is that, in a world

where we’re creating 2.5 quintillion bytes of data every day, not only is it expen-
sive to store old data, but the shelf-­life of data is shrinking. The most valu-
able insights are always going to be in the most up-­to-­date data. As Naveen
Peddamail, a Walmart senior analyst, told me, “If you can’t get insights until
you’ve analyzed your sales for a week or a month, then you’ve lost sales.”2

The exact data you need will be specific to your business, but as a starting
point (and this is something I talk about in my data strategy book), I always
advise business leaders to focus on the data that will solve your biggest busi-
ness problems and challenges, or help you answer your most pressing busi-
ness questions.

Proprietary data will set you apart – but don’t

overlook external data
You can harness data from both internal sources (proprietary data from
within the business) and external data from third-­party sources. Proprietary
data is obviously incredibly valuable because, by its very nature, only your
business has that information. That is the information that will set you apart
from your competitors.

That said, external data can be very useful. External data encompasses all the
data out there that isn’t owned by your business. Could be social media data.
Could be weather data or satellite data. Could be demographic data. Could be
internet search data. Wherever it comes from, you can glean valuable insights
from third-­party datasets. As an example, I know of one construction ­company
that uses high-­res satellite images which, combined with AI ­analytics, means
they can monitor progress on competitors’ construction sites.

Of course, there are challenges to working with external data. There may be
a cost to access that data. You don’t have direct control of the data (and you

234
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

may find yourself overly reliant on the data provider). You could lose access
to the data. And you’ll need to be sure that the data you’re buying or access-
ing has been collected and processed in a lawful, ethical way. But all things
considered, external data has the potential to be extremely rewarding – and
when combined with your precious internal data, allows you to build a wider
picture of what’s going on.

I’m aware that talk of data can make people seize up in anxiety, so to sum up
my tips for finding the right data: make sure you have a thorough, up-­to-­date
data strategy in place (one that’s been updated to take account of GenAI);
focus on the data that will solve your biggest business problems; think about
the shelf-­life of that data, and how you can get as close to real-­time data as
possible; and finally, start with your precious internal data, but also consider
the value of external data.

Getting the Right Technology in Place

Finally, having considered culture, skills, and data, you can then begin to
think about the technology you need to make all this happen. Again, every
business will have different technology needs, but broadly speaking, you can
think of technology requirements for GenAI in terms of two layers:

• The base layer is fast networks and connectivity, data infrastructure,

and cybersecurity. These three are the foundational tech building
blocks that will help you leverage GenAI (and other cutting-­edge tech-
nologies) successfully.
• On top of this sits your GenAI tools.

The foundation: networks, data infrastructure, and security

Before making any decisions about GenAI technologies, it’s really important
to get your foundational technologies right. These technologies aren’t nice-­
to-­have options – they’re absolute bedrock technologies. Must-­haves, for any
business, regardless of size or budget.

In terms of networks, you need fast, secure networks and easy connectiv-
ity, not just in your offices, but also on the go. The latest iteration in mobile

235
 GENERATIVE AI IN PRACTICE

network technology, 5G, will transform connectivity for many businesses,

bringing swift connections and real-­time collaborations, wherever your
people are.

In terms of data management, we’re not talking about collecting vast quanti-
ties of data – we’re talking about collecting the right data, at the right time,
and having the means to store, access, and use that data. There are lots of
options for data storage but in my opinion, cloud storage is the best option for
most businesses.

And in terms of cybersecurity, we’re talking about technologies that keep your
business safe from threats like ransomware, phishing, and breaches. Not only
does this involve deploying advanced threat detection mechanisms, it also
involves being aware of emerging threat landscapes, and training the work-
force so they can detect potential threats.

These three foundational elements are essential to businesses of all sizes and
budgets. A small company operating on a tight budget will obviously want to
keep the foundations simple, but they’re still necessary. If I were a small busi-
ness, I’d focus on beefing up network connectivity to ensure seamless opera-
tions, getting a decent cloud-­based data infrastructure in order, and ensuring
the best cybersecurity possible. Those really are the three basic things you can
do to prepare your business for next-­level technologies like GenAI.

If you have more money to throw at your foundational technologies, you

might consider leveling up your foundational elements. Choosing 5G for
faster, more robust connections, for instance. Or integrating paid-­for external
datasets to enrich your insights. Or even dabbling in synthetic data to expand
your repository.

Choosing GenAI tools

We’ve talked about GenAI tools throughout the book, and in the Appen-
dix you’ll find a handy list in one place. You can tap into readily available
tools like ChatGPT or its underlying GPT-­4 language model, but you’ll have
to consider the potential privacy implications. (At the very least, always

236
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

be aware of whether the information you enter will be used for training
­language models.)

Lots of organizations that I work with choose to create their own secure ver-
sion of GPT-­4, but that may not be feasible depending on your budget and
expertise. The great thing about GenAI is there are so many options, from
companies on a shoestring to companies with deep, deep pockets. If you don’t
have the in-­house knowledge to help you decide which tool or tools are right
for your business, it’s well worth consulting with an AI expert.

Remember to Take a Strategic Approach

Finally, to revisit what I said in Chapter 3, remember that GenAI has big
implications for businesses. As such, you will need to review your overarch-
ing business strategy to ensure it’s still relevant, and update it in line with the
incredible possibilities of GenAI.

When I work with companies on their AI strategy, what often happens is they
realize their existing business strategy is not fit for purpose. Because AI in
general – and specifically GenAI – will change their business, for example,
by rendering aspects of their service obsolete. So as a final reminder, please
do take a thorough look at your business strategy and see whether it is still
relevant in terms of where the world is heading.

When you’re happy that your business strategy is up to date, then you can cre-
ate an AI strategy, if you haven’t done so already. (And, again, if you do have
an existing AI strategy, it will also need to be reviewed and updated in line
with GenAI.) As you might expect, I have a book to help with AI strategy, too.
It’s called The Intelligence Revolution: Transforming Your Business With AI,
and it’s designed to help businesses plan for and implement AI successfully.

But at heart, an AI strategy starts with two simple steps. First, you identify
the potential applications of AI/GenAI in your business. These are your use
cases. Then you whittle those potential use cases down to just a few (one
to three) top-­priority use cases, plus one or two “quick-­win” uses. So, you’re
looking for a combination of high-­value, long-­term use cases that are key to

237
 GENERATIVE AI IN PRACTICE

long-­term business success, plus a couple of quick-­win projects that may not
be as strategically important, but help you experiment with AI/GenAI, build
confidence, and build buy-­in. Having identified your use cases, you can then
think about the data you need for those use cases, plus technology, and any
governance issues.

While the allure of exciting new technologies is strong, it’s really important to
take a strategic approach. In terms of GenAI, this starts with understanding
the impact of GenAI on your business operations, your products and services,
and maybe even your business model – and then updating your business
strategy accordingly, before you tackle an AI strategy and implementation.

Key Takeaways
We’ve covered a lot of ground in this chapter. To recap:

• Remember that GenAI is a tool. It won’t do our jobs for us, but we’ll use
it to do our jobs more effectively.
• Adopting GenAI successfully requires a shift in culture and mindset.
It requires a mindset that embraces curiosity, humility, adaptability,
and collaboration. It requires an organizational culture where people
continually challenge the status quo, are comfortable with change (and
failure), are not afraid to experiment, and are open to learning new
things. This culture and mindset must be modeled from the top down.
• To help with culture and mindset (and implementation in gen-
eral), I strongly recommend organizations appoint a chief AI officer
(CAIO) – a board-­level position that promotes a better awareness of AI
and ensures AI is being used effectively. As an alternative, you might
appoint an AI expert as a non-­executive director.
• In terms of skills and talent, AI delegation will become a crucial skill.
For leaders, managers, and individuals, this means working out what
we still need to do for ourselves, and what’s best left to machines. Tap-
ping into AI talent can be difficult for many businesses, so upskilling
your existing people is likely to be the best approach.

238
 IMPLEMENTING GENERATIVE AI: KEYS FOR SUCCESS

• AI is nothing without data, so if you don’t already have a data strategy

in place, now is the time to create one. With so much data on offer,
I recommend focusing on the data that will solve your biggest busi-
ness problems. Also, think about the shelf-­life of that data (real time or
near–real time is best), and consider both internal and external data.
• In terms of technology, there are three fundamental elements that will
help you leverage new technologies successfully: fast networks and con-
nectivity, data infrastructure, and cybersecurity. Your GenAI tools then
sit atop that bedrock. Turn to the Appendix for a list of GenAI tools.
• And finally, it’s really important to take a strategic approach. You need
to understand the impact of GenAI on your business operations, your
products and services, and maybe even your business model – and then
update your business strategy accordingly.

We’re almost at the end of our journey. In the final chapter, we’ll explore some
predictions for the future of GenAI. Some of which may surprise you. . .

Notes
1. Generative AI: The Mindset Divide That Will Determine Your Success; Forbes;
https://www.forbes.com/sites/bernardmarr/2023/11/13/generative-­ai-­t he-­
mindset-­divide-­that-­will-­determine-­your-­success/
2. How to Use Real-­Time Data: Key Examples and Use Cases; Bernard Marr;
https://bernardmarr.com/how-­t o-­u se-­real-­t ime-­d ata-­k ey-­e xamples-­a nd-
­use-­cases/

239
 19
GLIMPSES OF THE
FUTURE: PREDICTING
THE TRAJECTORY
OF GENERATIVE AI

Imagine a future where AI isn’t just smart but also your creative sidekick.
­Generative AI could turn brainstorming sessions into a wild ride of endless ideas,
design personalized virtual experiences, and maybe even help you write that
novel you’ve been dreaming about. It’s like having a super-­smart, ultra-­creative
companion at your fingertips, making the future look pretty darn exciting!

That’s what ChatGPT had to say when I asked it about the future of GenAI.
As for me, I have a few predictions of my own on how GenAI will evolve. Let’s
get into them.

Are We Moving Closer to General AI?

All of the GenAI capabilities that we’ve talked about so far in this book fall
under the umbrella of narrow or applied AI – meaning they simulate human
thought in order to carry out a specific task or function (like drafting an
email, or generating an image). We’re not yet at the point of general or gener-
alized AI – intelligent machines that can turn their attention to pretty much
any task, just like the human brain can. In other words, while today’s AIs do
a good job of simulating human thought, creativity, simple decision-­making,

241
 GENERATIVE AI IN PRACTICE

and conversation, they’re not able to fully simulate everything the human
brain does. But are we nearing that point?

Certainly, creativity (or the simulation of creativity) feels like a major mile-
stone on that journey. Something that was previously considered uniquely
human can now be done by machines to a pretty impressive standard. And
they will only get better and better. There’s no rolling back the clock on this
intelligence revolution. From here, there is only forwards; forwards toward
more intelligent machines.

Whether we’ll achieve this vision of truly intelligent machines is still up

for debate, but it does seem we’re edging closer to that point. We now have
machines that can see, hear, speak, read, write, and create. And increasingly,
GenAIs are gaining the ability to do many of these things at once – such as
being able to create text and images together. As an example, the third itera-
tion of text-­to-­image tool Dall-­E is able to generate high-­quality text embed-
ded in its images, putting it ahead of rival image-­generator tools.1 Imagine,
for instance, an AI-­generated image of a billboard sign with text on it, or a
picture of a brick wall plastered with graffiti text. Creating images like these,
with text in, has previously proven tricky for GenAIs. The most recent ver-
sions of GenAI tools can now see, hear, and speak, as well as write.2

So, one of my predictions is that GenAIs will continue this move toward mul-
timodal AIs that can create in multiple ways – and in real time, just like the
human brain. That will bring us another step closer toward general AI, or true
machine intelligence.

According to DeepMind co-­ founder, Mustafa Suleyman, the next step

beyond GenAI is “interactive AI” – by which he means bots that don’t just
chat, but can carry out all sorts of tasks by delegating to other software (and
even people) to get stuff done for you.3 In Chapter 16, we saw how GenAI
can be used to write or review computer code, meaning you could use it to
build, for example, a new recipe app designed around healthy eating. As AI
becomes more interactive, you could, in theory, task an AI with the entire
app-­creation project – from writing and testing the code, to hiring nutrition-
ists, food photographers, and recipe developers (some of whom may be bots),
to liaising with beta testers, to releasing the app, and organizing the marketing

242
 GLIMPSES OF THE FUTURE: PREDICTING THE TRAJECTORY OF GENERATIVE AI

campaign. In a simpler example, you could ask a bot, like ChatGPT, to book
a night away for you and your partner – including finding and booking the
hotel, choosing a restaurant for you, booking a table (and informing the res-
taurant of your partner’s egg allergy), and booking a walking tour of the city.
You could delegate the entire task to a bot.

This takes the notion of multi-­model, real-­time GenAIs to a whole new level,
doesn’t it? If Suleyman is right, AIs will become much more interactive and
helpful – making our lives easier every day by taking on the burden of mun-
dane tasks and allowing us to focus on . . . well, who knows what? Organiz-
ing our sock drawers, perhaps. The point is, it’s another step on the journey
toward generalized AI – intelligent machines that can do anything the human
brain can do.

But how else can we expect GenAI to evolve? I have more predictions up my
sleeve. . .

Combining Generative AI with Robots

If you’ve ever seen any of Boston Dynamic’s videos, you’ll know that robots
are developing at an impressive rate. In the not-­too-­distant future, I believe
we’ll see robots being routinely equipped with GenAI capabilities – thereby
vastly increasing the range of tasks that both robots and AIs can take on for us.

So far, GenAI enables us to automate and augment cognitive and creative jobs,
including the work of doctors, designers, musicians, marketers, and more.
Physical jobs such as building, assembly line work, cleaning, and so on are
(by and large) unaffected by the wave of GenAI transformation. But combine
GenAI with robots and that may change. We could see AIs working in all sorts
of sectors. On factory floors, checking in on patients in hospital wards, on con-
struction sites, in retail stores, in hotels, in restaurants, you name it. By embed-
ding GenAI in robots, you have machines that can not only perform tasks but
also generate new customer experiences. For instance, a hotel concierge robot
could tailor travel itineraries in real time, based on the guests’ preferences.

Looking further ahead, we may even live with intelligent robots in our homes.
Imagine having an all-­knowing robot companion in your home – a robot

243
 GENERATIVE AI IN PRACTICE

powered by GenAI that could answer any question you have, while also
­hoovering the floor, managing your schedule, planning what to have for din-
ner, helping you write that report for work, creating personalized learning
plans for your children, and even suggesting what movie to watch in the even-
ing based on your mood. It’s like having an Alexa, Roomba, cleaner, personal
chef, personal assistant, tutor, and babysitter all rolled into one.

NVIDIA is already working on combining GenAI with robotics. The com-

pany has updated its Jetson AI and robotics platform – which has long been
adopted across industries including manufacturing and transportation – to
facilitate the use of GenAI.4 The main idea here is that GenAI brings advance-
ments in computer vision (the ability for machines to “see”), by giving
machines the ability to learn faster and become more accurate at identify-
ing and reacting to the things around them. Traditionally, computer vision
relies on systems being trained on vast amounts of data, with the machine
learning to recognize specific objects and then react in specific ways. When
a system comes across an object it doesn’t recognize, it’s stumped. Which is
where GenAI comes in. GenAI offers greater generalization, because it can
recognize and interact with elements it hasn’t necessarily been trained on.

For industries that use robots relying on computer vision, GenAI could prove
transformative. But it will also transform robots in other ways. For example, it
could give machines the ability to communicate with colleagues around them.

In other words, although we may be some way away from intelligent home
robots, in the very near future, GenAI will begin to impact robots in indus-
trial and other work settings.

Brain–Computer Interfaces Will Allow for More

Intuitive Interactions
At the moment, you need to go to an app or web page to interact with GenAI,
but how about having an almost direct link to a GenAI system from your
brain? You simply speak a question or request, and the GenAI responds. This
could be achieved through smart glasses or smart contact lenses, or even
through implants in the brain. Whatever the hardware looks like, the point

244
 GLIMPSES OF THE FUTURE: PREDICTING THE TRAJECTORY OF GENERATIVE AI

is: the way in which we interact with GenAI may evolve to become much
more intuitive and instant.

If it sounds like I’ve gone off the deep end, hold on. Because research is already
well underway into the field of brain–computer interfaces (BCIs) – means for
us to interact with machines or the internet through thoughts alone. Boston-­
based start-­up, Neurable, is already working in this field, creating sensors that
can decipher brain activity, understand the user’s intention, and then translate
that into virtual reality. Elsewhere, Elon Musk–­founded Neuralink is devel-
oping implants that will allow two-­way communication between the human
brain and an app. Meta is also developing its own BCIs that can decode speech
directly from the brain.

Developments in BCIs could end up transforming humans into practically

a new species. To some extent this merging of humans and machines is
already happening. People are being fitted with robotic arms and bionic eyes,
for instance. So why not augment the human brain with the ability to tap
directly into AI?

Nobody knows how the field of BCIs will develop, but I think it’s a real pos-
sibility that we could, in future, be collaborating with machines in more and
more intuitive ways.

Imagine the implications for human creativity and efficiency, if BCIs enable
direct communication between the human brain and GenAI systems. Ideas
could be generated, refined, and brought to fruition with unprecedented speed
and precision. Decision-­making and innovation would be vastly improved if
we could combine human cognition and experience with the data-­processing
prowess of AI. It’s a really interesting prospect, and – I admit – a somewhat
wild prediction for the future. But with the speed at which AI and BCIs are
developing? It’s certainly not impossible.

Using Generative AI for a Better World

The final point I want to leave you with is this: as one of the most transforma-
tive technologies we’ve ever seen, GenAI can be a force for enormous good
in this world.

245
 GENERATIVE AI IN PRACTICE

GenAI can help us tackle some of the biggest problems on the planet. Yes,
GenAI can be used to draft a funny speech for your friend’s wedding, or come
up with an idea for tonight’s supper. But it can also be used to address issues
like climate change, massive inequality, hunger, access to healthcare, and more.
It can help us find new treatments for diseases. It can democratize access to
mental health support. It can predict how infectious diseases will progress. It
can address food security through crop yield optimization. It can do so much
good. When you think about it, it’s rather like giving humans superpowers.

Of course, on the flipside, GenAI can also be used to spread misinformation,

create new biological weapons, or even build autonomous weapons that can
identify and neutralize threats without any human oversight.

Which is why we need to ensure we use this super-­potent technology for the
right things. We will need regulations that will protect us and ensure GenAI is
used for positive transformation – and not to create more polarization, med-
dle in elections, deliberately spread false information, and the like. We will
need transparency around how these systems are used. We will need ethical
guidelines and frameworks in place.

To some extent this is beginning to happen. Meta, for example, has announced
that political ads running on Facebook and Instagram will be required to dis-
close if they contain AI-­generated content and images.5 It’s an attempt to help
voters make more informed decisions about the content they see online in
the run-­up to elections. Other tech companies are following suit. Google, for
example, has a similar labeling policy, requiring political ads on YouTube and
other Google Platforms to disclose the use of AI-­altered voices or imagery.6
Efforts like these will help to combat GenAI’s ability to create realistic-­looking
(but entirely fake) videos, images, and audio.

We will also need technology companies (and organizations implementing

GenAI tools) to address limitations around bias and hallucinations. Navigat-
ing a problem like hallucinations is tricky when it’s still unclear why GenAIs
occasionally say the wrong thing. But there are workarounds, such as running
two GenAI models in parallel, with one model fact-­checking the other. The
good news is, a lot of work is going into this area, as it’s in the interests of all
major stakeholders in the industry to solve these problems.

246
 GLIMPSES OF THE FUTURE: PREDICTING THE TRAJECTORY OF GENERATIVE AI

So yes, there are challenges to overcome, and it would be naive to imagine

GenAI will only be used for good. But my hope is GenAI will be overwhelm-
ingly used for good, to tackle some of humanity’s most pressing challenges
and make our world better.

This is our opportunity. And what an opportunity it is.

Connect with Me
That brings us to the end of my predictions. But what do you think? Are you
excited by the idea of GenAI or a little weirded out by it?

One of the things I love about writing books like this is it’s a springboard
for further discussion. So please do feel free to ask questions or share your
thoughts. How are you beginning to use GenAI in your everyday life and
work? How might GenAI help you solve your organization’s biggest prob-
lems? How might your workplace be transformed by GenAI and other future
technologies? What are your biggest implementation challenges? How do you
see your own job role evolving?

And of course, you can always get in touch if you need help planning for this
AI-­driven revolution and implementing future technologies. I consult with
businesses of all shapes and sizes.

You can connect with me on the following platforms:

LinkedIn: Bernard Marr

Twitter: @bernardmarr
YouTube: Bernard Marr
Instagram: @bernardmarr
Facebook: facebook.com/BernardWMarr

Or head to my website at www.bernardmarr.com for more content (including

my podcast), and to join my weekly newsletter, in which I share the very latest
information.

247
 GENERATIVE AI IN PRACTICE

Notes
1. DALL-­E 3 could take AI image generation to the next level; Digital Trends;
https://www.digitaltrends.com/computing/dall-­e-­3-­testing-­leaked-­revealing-­
interesting-­new-­text-­features/
2. The New ChatGPT Can ‘See’ and ‘Talk.’ Here’s What It’s Like; NY Times; https://
www.nytimes.com/2023/09/27/technology/new-­chatgpt-­can-­see-­hear.html
3. DeepMind’s co-­founder: Generative AI is just a phase. What’s next is interactive AI;
Technology Review; https://www.technologyreview.com/2023/09/15/1079624/
deepmind-­inflection-­generative-­ai-­whats-­next-­mustafa-­suleyman/
4. Generative AI Is Coming to Robots, Courtesy of NVIDIA; eWeek; https://www
.eweek.com/artificial-­intelligence/generative-­ai-­robots-­nvidia/
5. To help 2024 voters, Meta says it will begin labeling political ads that use AI-­
generated imagery; AP News; https://apnews.com/article/meta-­facebook-­
instagram-­political-­ads-­deepfakes-­2024-­c4aec653d5043a09b1c78b4fb5dcd79b
6. To help 2024 voters, Meta says it will begin labeling political ads that use AI-­
generated imagery; AP News; https://apnews.com/article/meta-­facebook-­
instagram-­political-­ads-­deepfakes-­2024-­c4aec653d5043a09b1c78b4fb5dcd79b

248
 APPENDIX: GENAI TOOLS

MULTIMODAL MODELS

ChatGPT
The application that kicked off the current generative AI craze, ChatGPT,
has evolved since it emerged in 2022 and now includes multimodal capabili-
ties, as well as access to GPT-­4, the most powerful publicly available language
model. It could fit into many of the categories in this list, so I’ll just put it here!

https://chat.openai.com/

Google Bard
Bard is a multimodal chatbot much like ChatGPT but Google users might
appreciate its extensive integration with tools like Gmail, Drive, and Maps.

Google Vertex AI
Multimodal search combining language and vision models aimed at enabling
businesses to organize and analyze data.

https://cloud.google.com/blog/products/ai-­machine-­learning/multimodal-­
generative-­ai-­search

249
 APPENDIX: GenAI TOOLS

Meta Imagebind
Provides a multimodal platform based around six “sensory” datatypes –
images, text, audio, depth, thermal, and inertial, enabling it to learn holisti-
cally, much like we do.

https://imagebind.metademolab.com/

Samsung Gauss
Samsung has created a multimodal suite of generative AI tools that will be
integrated across its line of devices.

Writing/ Text Generation Tools

AI Writer
Writes content like blog posts while citing its sources, to help overcome issues
of AI hallucination.

https://ai-­writer.com

Anyword
Marketing-­focused content generation tool with predictive performance
scoring and audience analytics.

https://anyword.com

Copy.ai
Content creation tool aimed at businesses and enterprises, with built-­in
­templates for common jobs.

https://www.copy.ai

250
 APPENDIX: GenAI TOOLS

Coursera
The online learning platform has created generative AI tools for students as
well as teachers and course authors.

https://www.coursera.org/

DuoLingo
One of the most popular language-­learning platforms, now with generative
AI roleplay conversations and personalized feedback powered by GPT-­4.

https://www.duolingo.com/

Jasper
Content creation tool aimed at marketers wanting to automate copywriting
tasks like blogs, product descriptions, and social media posts.

https://www.jasper.ai/

Lex
An AI co-­pilot for writers offering AI feedback on your work and tools for
finding the right words or expressions, or generating ideas.

https://lex.page

Rytr
Generates blogs, emails, social media posts, SEO headlines, and ads.

https://rytr.me

251
 APPENDIX: GenAI TOOLS

Scribe
Automates the creation of documentation, step-­by-­step and how-­to guides.

https://scribehow.com/tools/ai-­text-­generator

Wordtune
Generative writing assistant aimed at professionals.

https://www.wordtune.com/

WriteSonic
Geared toward creating SEO content and traffic-­boosting copy.

https://writesonic.com

Education Tools

Cognii
Education platform offering personalized tutoring and student assessments
generated by AI.

https://www.cognii.com/

Fetchy
Help with writing, planning, organizing, and lesson prep for teachers.

https://www.fetchy.com/

252
 APPENDIX: GenAI TOOLS

Gradescope
AI-­augmented grading platform for educators; also enables peer evalua-
tion of work.

https://www.gradescope.com/

Ivy
An AI framework for creating customer service chatbots, with a focus on
helping institutions of higher learning.

https://ivy.ai/

Knowji
AI-­powered language learning that creates customized lessons.

https://www.knowji.com/

Plaito
Generative text tool designed to help writers improve their skills with person-
alized feedback.

https://www.plaito.ai/

Querium
Personalized, step-­by-­step tutoring in STEM subjects.

https://www.querium.com/

253
 APPENDIX: GenAI TOOLS

Image, Art, and Design

Adobe Firefly
AI-­powered graphic design for creating any kind of visuals; integrates with
Adobe’s creativity tools.

https://www.adobe.com/uk/sensei/generative-­ai/firefly.html

Autodesk
This industry-­standard design platform now provides generative tools aimed
at improving users’ skills and boosting creativity.

https://www.autodesk.com/

Canva
Generative design tools are now baked into the user-­
friendly graphic
design platform.

https://www.canva.com/

Dall-­E
Creates images based on text descriptions, now included in ChatGPT Plus.

https://openai.com/dall-­e-­2

Durable
Generate entire websites along with images and copy from language prompts.

https://durable.ai

254
 APPENDIX: GenAI TOOLS

Midjourney
Images from text; very powerful but currently only accessible over Discord.

https://www.midjourney.com/

Stable Diffusion
One of the most popular image generation models, unlike Dall-­E it is
open source.

https://stablediffusionweb.com/

Video

Colossyan
Generate synthetic content tailored for workplace learning.

https://www.colossyan.com/

Deepbrain
AI-­generated virtual avatars and synthetic media aimed at corporate and
business users.

https://www.deepbrain.io/

Descript
All-­in-­one AI video-­editing platform with script-­based editing and auto-
mated transcription.

https://www.descript.com/

255
 APPENDIX: GenAI TOOLS

Designs.AI
Generate graphics, videos, logos, and other design assets.

https://designs.ai/

Elai
Generative video tool geared toward designing training videos.

https://elai.io/

Filmora 13
Full AI-­powered video-­editing suite with chatbot co-­pilot to help you out.

https://filmora.wondershare.com/

Fliki
Transform text into lifelike speech and videos, make videos in minutes.

https://fliki.ai/

HeyGen
Create talking photographs and other cool AI video tricks.

https://app.heygen.com/

Invideo
Create videos using a simple interface and pre-­made templates.

https://invideo.io/

256
 APPENDIX: GenAI TOOLS

OpusClip
Automagically transform long-­form videos into short, punchy viral clips.

https://www.opus.pro/

Peech
Create branded videos from your existing content library for podcasts, webi-
nars, or social media.

https://www.peech-­ai.com/

Pictory
AI video-­editing platform that creates, edits, and narrates videos.

https://pictory.ai

Runway
Video generation for creatives with a focus on storytelling.

https://runwayml.com/

Synthesia
Turn text into video and create virtual video avatars that synthesize speech in
multiple languages.

https://www.synthesia.io/home

257
 APPENDIX: GenAI TOOLS

Synthesys
Generate professional AI-­powered avatars and voiceovers.

https://synthesys.io/

Type Studio
Create and edit videos simply by editing transcribed text.

https://topai.tools/t/type-­studio

VEED:IO
Generative AI video creation and editing.

https://www.veed.io/

Sound, Music, and Voice Generators

Amadeus Code
Generative AI-­powered songwriting assistant that allows users to pay per fin-
ished track.

https://amadeuscode.com/

AIVA
Great for those wanting to use AI to develop complex and emotional music
pieces that sound like they were created by human composers.

https://www.aiva.ai/

258
 APPENDIX: GenAI TOOLS

Beatoven
Generate background music for online content (or any other type of music) in
multiple styles, and edit with simple AI tools.

https://www.beatoven.ai/

Boomy
Create songs in seconds with a simple interface and a strong user community.

https://boomy.com/

Loudly
Music platform for creating AI-­
generated, royalty-­
free tracks, with AI-­
assisted recommendations.

https://www.loudly.com/

Mubert
Synthetic music generator that creates tunes to fit the mood of your content.

https://mubert.com/

Murf
AI voice studio with realistic and customizable text-­to-­speech.

https://murf.ai/

259
 APPENDIX: GenAI TOOLS

MusicLM
A Google AI Test Kitchen project that users can sign up to try out – ­Generative
music creation from Google.

https://aitestkitchen.withgoogle.com/experiments/music-­lm

Soundful
Generate unique tunes with the help of AI at the click of a button.

https://soundful.com/

Speechify
Text-­to-­speech tool for creating natural sounding synthetic voice.

https://speechify.com/

Soundraw
Create custom music tracks in many different styles and moods, for
royalty-­free use.

https://soundraw.io/

Splash
Create music based on text prompts, including vocals.

https://pro.splashmusic.com/

260
 APPENDIX: GenAI TOOLS

Wavtool
Music generator with user-­friendly Conductor tool that guides even complete
beginners through the process of creating AI music.

https://wavtool.com/

Coding Tools

AIXCoder
Augments programming abilities by providing intelligent code completion
and optimization.

https://www.aixcoder.com/

Amazon CodeWhisperer
Automated code builder and assistant from Amazon, with real-­time help
using natural language prompts.

https://aws.amazon.com/codewhisperer/

Codacy
Automated code review and debugging.

https://www.codacy.com/

Codecomplete
Edit, analyze, and explain code in real time in order to streamline soft-
ware creation.

https://codecomplete.ai/

261
 APPENDIX: GenAI TOOLS

Code Llama
Meta’s AI coding assistant built on a tuned version of its Llama 2 language
model, specializing in creating, analyzing, and interpreting code.

https://about.fb.com/news/2023/08/code-­llama-­ai-­for-­coding/

Hugging Face
Hugging Face is actually a community and collaborative platform, but it pro-
vides access to a large number of generative tools, models, and datasets.

https://huggingface.co/

Github Copilot
Provides contextually relevant code suggestions for programmers using OpenAI’s
Codex model, a version of its popular GPT-­3 fine-­tuned for coding tasks.

https://resources.github.com/copilot-­for-­business/

IBM WatsonX Code Assistant

Generative AI coding assistant from IBM powered by its Granite foundation
model, with a focus on creating Enterprise-­grade AI.

https://www.ibm.com/products/watsonx-­code-­assistant

Metabob
Automated code review that analyzes code for errors or bugs and sug-
gests fixes.

https://metabob.com/

262
 APPENDIX: GenAI TOOLS

Mintlify
Automates the creation of programming documentation.

https://mintlify.com/

Mutable
AI-­accelerated software development platform with automated test generation.

https://mutable.ai/

Replit
A cloud-­based integrated development environment (IDE) augmented with
generative functionality, and collaborative features.

https://replit.com/

Studio Bot
A generative AI tool from Google designed to help with software develop-
ment for its Android platform.

https://developer.android.com/studio/preview/studio-­bot

Tabnine AI Code Assistant

AI assistant for software development with a focus on security.

https://www.tabnine.com/ai-­assistant-­for-­software-­development-­r

263
 APPENDIX: GenAI TOOLS

Warp
Not a code writer, but a terminal application that offers contextual autosug-
gestion, error correction, and suggestions on specific commands that a user
might want to use.

https://www.warp.dev/

Data Analytics

Akkio
Generative BI platform that enables predictive modeling and insight generation.

https://www.akkio.com/

Alteryx
The Alteryx platform now incorporates a no-­code AI Studio to let users ­create
their own analytics apps, using custom business data.

https://www.alteryx.com

Microsoft Power BI
Microsoft has added generative analytics capabilities to Power BI, utilizing its
Co-­Pilot and Fabric AI platforms.

https://powerbi.microsoft.com/

Microstrategy
Generative AI assistant provides quick answers and insights into business data.

https://www.microstrategy.com

264
 APPENDIX: GenAI TOOLS

Qlik
Another established analytics and data platform that now lets users embed
generative AI analytics content into their reports and dashboards.

https://www.qlik.com/

Tableau
Tableau Pulse helps users use generative AI to extract meaningful insights
from data.

https://www.tableau.com/

Synthetic Data

BizData X
Simplifies data masking and anonymization with synthetic data generation
for business.

https://bizdatax.com/

CTA Test Manager

Allows the creation of very technical and complex datasets.

https://www.broadcom.com/products/software/continuous-­testing/test-­
data-­manager

Cvedia
Computer vision and video analytics powered by synthetic data.

https://www.cvedia.com/

265
 APPENDIX: GenAI TOOLS

Datomize
Create datasets with dynamic validation tools to ensure they are as realistic
as possible.

https://www.datomize.com/

Edgecase
Create labeled synthetic data as-­a-­service.

https://www.edgecase.ai/

GenRocket
Dynamic data generation with enterprise scalability, for creating data, to
assist with software testing.

https://www.genrocket.com/

Gretel
Synthetic data platform targeted at software development.

https://gretel.ai/

Hazy
Generates synthetic data allowing customers to “mask” their real data, to
solve security and privacy challenges.

https://hazy.com/

266
 APPENDIX: GenAI TOOLS

K2View
Generates data for the purpose of training machine-­learning models.

https://www.k2view.com/solutions/synthetic-­data-­generation-­tools/

KopiKat
No-­code data augmentation designed to enhance privacy and improve per-
formance of neural networks.

https://www.kopikat.co/

MDClone
Synthetic data aimed at healthcare professionals.

https://www.mdclone.com/

Mostly
Generate synthetic data that mimics real-­world data.

https://mostly.ai/

Sogeti
Billed as a “data amplifier”; mimics real datasets by matching the characteris-
tics and correlations of existing data.

https://www.sogeti.com/services/artificial-­i ntelligence/artificial-­d ata-­

amplifier/

267
 APPENDIX: GenAI TOOLS

Syntho
Self-­service data generation for insights and decision-­making.

https://www.syntho.ai/

Tonic
The “fake data company” provides a comprehensive platform for developing
realistic, compliant, and secure synthetic data.

https://www.tonic.ai/

YData
Automated synthetic data generation to enhance productivity and AI model
performance.

https://ydata.ai/

Productivity Tools

Airgram
Automatically record, transcribe, summarize, and share meeting transcripts
and conversations.

https://www.airgram.io/

Beautiful
Transform ideas into fully realized slides and presentations with automated
design features.

https://www.beautiful.ai/

268
 APPENDIX: GenAI TOOLS

ClickUp
Generative AI-­driven workplace productivity suite with project management,
time tracking, whiteboarding, agendas, and other features.

https://clickup.com/

Decktopus
Generate slide decks and presentations from natural language prompts.

https://www.decktopus.com/

Duet
Another Google service, this one is built into Workspace to provide a collabo-
rative assistant for productivity tasks.

https://workspace.google.com/blog/product-­a nnouncements/duet-­
ai?ref=the-­writesonic-­blog-­making-­content-­your-­superpower

Fireflies
Transcribe, summarize, and search voice conversations.

https://fireflies.ai/

Notta
AI notetaking and audio transcription.

https://www.notta.ai/en

269
 APPENDIX: GenAI TOOLS

Cyber Security

Airgap
Comprehensive AI security-­as-­a-­service that allows users to explore threats
and analyze data with natural language processing.

https://airgap.io/

Crowdstrike Charlotte AI
Cloud security that you can talk to and question to get real-­time, understand-
able insights.

https://www.crowdstrike.com/products/charlotte-­ai/

Google Cloud Security Workbench

Google’s comprehensive cloud security services include a powerful LLM-­
driven threat intelligence tool.

https://cloud.google.com/security/ai

Microsoft Security Co-­Pilot

Summarizes threats and security signals with natural language generation.

https://www.microsoft.com/en-­us/security/business/ai-­machine-­learning/
microsoft-­security-­copilot

270
 INDEX

2D sketches, translation, 184 AIXCoder, 261

3D modeling, GenAI (impact), 12 Akkio (Generative Reports tool),
3D printing, 182–183 219–220, 264
GenAI, impact, 25 Alexa (Amazon), GenAI
growth, 126 (integration), 36–37
All England Club, The (AI usage), 82
A Allen & Overy, Harvey
Ablo (Space Runners creation), 126 partnership, 168–169
Academic research, GenAI AlphaCode AI (DeepMind), coding
assistance, 185–186 competition, 201
Accelerate (LinkedIn feature), 101 Altered AI voice library, usage, 158
Accountants, GenAI Alteryx, GenAI leverage, 219, 264
assistance, 192–193 Amadeus Code, 258
Accounting, GenAI (usage), 197 Amazon Web Services (AWS)
Ada (doctor-developed AI app), 144 cloud-based operations, renewable
Advertisements (creation) energy usage, 57
aspects, automation, 99–100 GenAI models (availability), 35–36
examples, 100–101 machine-learning technology
GenAI, usage, 99–102 usage, 114
Advertising Ambient Assist (NextGen Healthcare)
AI, usage, 97 notetaking tool, 148
content creation, streamlining, 38 Anomaly detection, GenAI
GenAI (impact), 13
advantages, 98 Anti-Money Laundering (AML),
application, 97–99 GenAI assistance, 192
Airgap, 270 Anyword, 250
Airgram, 268 App experience,
AIVA, 258 improvement, 122–123
Aiva (AI music generator), 88 Aprimo, Azure OpenAI/ChatGPT
AI Writer, 250 usage, 101

271
 Index

Architectural design, GenAI Augmented reality (AR) technology,

(impact), 13 usage, 123
Artbreeder (AI image Autodesk, 180–181, 254
generation tool), 92 Auto-GPT (coding tool), 207
Art, GenAI (impact), 90–92 Automation, impact, 66–67
Artificial intelligence (AI) Automotive industry, GenAI
AI-powered role-playing chatbot, examples, 183–184
Bloom launch, 34 Avatar Cloud Engine (ACE) For
artworks/image generators, Games (NVIDIA), 160–161
examples, 91–92 Azure OpenAI (voice-
bias/explainability, 51–52 generation tool), 89
concept, introduction, 17–19
defining, 3 B
delegation, ability, 230–231 Banking/financial services
detection, 47–48 AI, impact, 189
disruptive force, 189 automation, 190–192
environmental impact, 56–57 GenAI uses, 189–193
influencers/models, 103–105 Banking operations (streamlining),
jobs, creation, 73–74 GenAI usage, 195
“responsible guardrails” Banks (service improvement), GenAI
(establishment), 58–59 usage, 194–195
revolution, 201 Bard (coding tool), 207, 249
shopping assistants, 121 Beatbot (AI music generator), 88
talent Beatoven, 259
retention, 232–233 Beautiful, 268
upskilling/partnerships Be My Eyes, Open AI
(usage), 231–232 (collaboration), 34–35
voice generators, 88–89 Biased data, problem, 51
winters, 18 Big data, impact, 19
Artificial Intelligence Act (European Bing Search (Microsoft), ChatGPT
Parliament), 56 integration, 35, 39
Artists, GenAI usage, 91 Bishop, Diamond, 206
Associated Press (AP), BizData X, 265
GenAI usage, 80 Body Labs, mixed reality technology
Audiobooks, GenAI (usage), 83–85 usage, 124
Audio content, GenAI creation, 85 Books, GenAI (usage), 83–85
Augmend, GitHub Copilot tool Boomy, 259
(usage), 205–206 Botatouille (AI recipe generator), 80

272
 Index

Brain-computer interfaces, Cloud, carbon footprint, 57

26, 244–245 Cloud computing, 23
Brands, GenAI (leveraging), 101–102 Cloud Security Workbench
Brands, GenAI (usage), 98 (Google), 270
BT Group, GenAI capabilities, 112 Cloud Skills Boost platform
Business (Google), 232
ecosystems, transformation, 29 CoCounsel (CIP GenAI
models, reshaping, 32–33 platform), 167–168
processes, refinement, 36–39 Codacy coding tool, 207, 261
BuzzFeed, GenAI (usage), 80 Codecomplete, 261
Code, GenAI generation/
C review, 202–203
California Innocence Project Code Llama (coding tool), 207, 262
(CIP), 167–168 CodeWhisperer (Amazon coding
Call note automation tool tool), 207, 261
(Sprinklr), 111 CodeWP (coding tool), 207
Candy (Prada virtual muse), 104 Codey (coding tool), 207–208
Canva, 254 Coding
Chat Generative Pre-trained AI, impact, 201
Transformer (ChatGPT), tools, examples, 205–209
3–4, 207, 249 Cognii (education tool), 252
ChatGPT Plus, 33 Colossyan, 255
Coca-Cola usage, 100 Computer code/coding, GenAI
freemium model, 32–33 generation, 9, 14
GenAI, contrast, 8–9 Computing technology,
generation capabilities, 52 advances, 23–24
information delivery, 70 Consensus (academic answer
marketing usage, 38 provider), 186
release, 21–22, 27 Content
tools, 7, 14, 18, 54, 121 creation, streamlining, 38
usage, 159 farms, fake content creation, 46–47
CheftGPT, usage, 42 GenAI creation, 5
Chest radiographs, issues personalization, GenAI
(detection), 145 (usage), 102–103
Clickbait articles, 47 presentation, 79–80
ClickUp, 269 Copilot (Microsoft), 218
Clinicians, information search Copy.ai (content creation
improvement, 149 tool), 38, 250

273
 Index

Copyright, GenAI issues, 52–54 infrastructure, 235–236

Coursera (online learning), 232, 251 literacy, importance, 212–213
Credit decisioning, GenAI optimization, 110
(usage), 190 overwhelm, problem, 211–212
Crowdstrike Charlotte AI, 270 preparation, 215
CTA Test Manager, 265 privacy/security, GenAI
Customer issues, 54–55
conversational search functions, problem, defining, 211–217
providing, 110 security, 235–236
engagement (reinvention), understanding, GenAI
intelligent systems (usage), 109 ability, 213–214
insights, GenAI (impact), 14 value, 233–235
journey, personalization, 125–128 Data Strategy (Marr), 233
personalized recommendations, Datomize, 266
giving, 110 Daugherty, Pal R., 111
Customer service Day a Computer Writes a Novel, The
enhancement, GenAI (usage), 116 (GenAI creation), 83
examples, 111–112 Decision Dilemma (decision paralysis)
GenAI, usage, 39, 109–112, 190 (data anxiety), 211–212
human customer service agents, Decktopus, 269
support, 110 Decorify (Wayfair GenAI feature), 124
jobs, loss (potential), 66–67 Deep Angel (AI image
Cvedia, 265 generation tool), 92
Cybersecurity threats, 193 DeepArt.io (AI image
generation tool), 92
D Deepbrain, 255
Dadabots (AI music generator), 88 Deep Dream Generator (AI image
Dall-E 2 (OpenAI), 4, 22, 38, generation tool), 92
92, 158, 254 Deepfakes, problems, 9, 46–47
watermark, usage, 48 Deep learning, advancement, 19–20
Data DeepMind (Google), 178, 242
analysis, GenAI generation, coding competition, 201
9, 13–14 Descript, 255
analysts (work loss), GenAI Design
(impact), 216 AI, usage, 173
analytics Autodesk research lab,
automation, 215 impact, 180–181
tools, 217–220 GenAI-powered design, business
augmentation, GenAI (impact), 14 argument, 181–182
examples, 217–220 generative design, 9, 12–13, 178–184

274
 Index

Designs.AI, 256 E
Development, AI (usage), 173 eBay
Devices, GenAI (incorporation/ magical listing tool, 129
examples), 115–118 ShopBot, 127
Diffusion models, function, 8 Edgecase, 266
Digital shopping experiences, Edge computing, 23
improvement, 121–124 Education, future, 133
Digital twins, GenAI (impact), 185 Educators, value (GenAI
Dillon, Jeff, 33–34 addition), 134–135
Dimitrov, Georgi, 33–34 Einstein GPT (Salesforce),
Discriminative AI, 4 101, 218–219
Diseases Elai, 256
detection, 143–145 Elf Tech (AI-generated software),
image analysis, 144–145 89
Disinformation, potential, 46–47 Elicit (academic paper search
DNA synthesis (gene synthesis), tool), 186
GenAI role, 178 ELIZA (AI program), 18
Doctors, precision care Ellison, Larry, 206
(providing), 146–147 Enhanced editing, GenAI (impact),
Domain-specific generative models, 22 11
Dotcom revolution, 19 Entertainment, transformation, 79
DreamGF users, characteristics Ernie (Baidu ChatGPT), 39–40
selection, 33–34 Etcembly, GenAI usage, 177
Drug discovery Expanding Agent Replies
GenAI, impact, 15, 173–178 (Zendesk), 111
phases, 175–176 Expedia Group, open-ended chat
Drugs (usage), 122
research, Google DeepMind Expert systems, growth, 18
support, 178 Explainable AI (XAI), 51–52
search (acceleration), GenAI External data, importance,
(usage), 174–176 234–235
Dubai Electricity and Water Authority
(DEWA), GitHub Copilot F
(usage), 206 Fake content, creation, 46
Dubbing (improvement), GenAI False information, GenAI
(usage), 87 involvement, 9
Duet, 269 Fan engagement (enhancement),
Duolilngo, 251 GenAI (usage), 81–83
OpenAI (collaboration), 136 Fashion design, GenAI (impact), 13
Durable, 254 Fast fashion, GenAI (usage), 183

275
 Index

Fetchy (education tool), 252 Generative adversarial networks

Film, GenAI (examples), 86–87 (GANs), 7, 20
Film, GenAI (usage), 85–87 Generative AI (GenAI)
Filmmakers, GenAI usage, 86 abilities, 8–15
Filmora 13, 256 advancement, 20–23
Financial modeling, GenAI advertisements, examples, 100–101
(usage), 191 artistic promise, 90
Financial professionals, GenAI capabilities, 110–111, 202–203
assistance, 195–196 challenges, management, 45
Financial sector, GenAI ChatGPT, contrast, 8–9
examples, 194–196 control, loss (reaction), 49–51
Fireflies, 269 criminal applications, 9
Firefly GenAI (Adobe), 99–100, 254 culture/mindset, fostering, 226–230
Flawless, GenAI usage, 87 cybersecurity threats, 193
fleetAI (Dentons ChatGPT tool), 168 daily usage, 39–42
Fliki, 256 data privacy/security issues, 54–55
Forecasting, GenAI (impact), 14 defining, 3–5
Formula One (F1), AI usage, 82–83 dependence/skills degradation, 49
Foundation models, usage, 8 ethical/societal concerns, 45–50
Fox Sports, Google (GenAI evolutionary blueprint, 17
collaboration), 83 examples, 4
Fox TV, Waymark financial stability risk, 193
(partnership), 101–102 foundations, 19–20
Francis, Jon, 206 function, explanation, 5–8
Fraud detection frameworks, GenAI GenAI-powered analytics
(usage), 191 tools, 218–220
Freddy AI bot (PhonePe usage), 112 hardware, environmental impact, 57
Freemium model, 32–33 healthcare promise, 150
Freshworks, ChatGPT usage, 205 impact, 16, 185
Frey, Bridget, 206 implementation/usage,
Future Internet, The (Marr), 90 225, 245–247
Future Skills (Marr), 137 infusion, 35–36
jobs, creation, 73–74
G journalism, relationship, 79–81
Gauss (Samsung), 150 legal concerns, 52–56
Geek Way, The (McAfee), 227 limitations, 216–217
GenAI chatbot, Babylon Health Meta (Facebook) usage, 37
development, 30 metaverse, relationship, 24–25
General AI (generalized AI), 241–242 models (black box perception), 51

276
 Index

multimodal models, 249 Graphic design, GenAI

performance risk, 193 generation, 12–13
power, harnessing, 211 Greifeneder, Bernd, 203
regulatory catch-up, 56–57 Gretel, 266
risks, 193
management, 45 H
robots, combination, 243–244 Haley, Mike, 179–181
self-regulation, importance, 58–59 Hazy, 266
skills/talent, cultivation, 230–233 Healthcare
strategic approach, 237–238 administrative improvements,
technology delivery, 147–149
advances, link, 23–26 crisis (solving), GenAI
risk, 193 (usage), 149–151
tools, selection, 236–237, 249 drugs, research an d
traditional AI, contrast, 4–5 development, 147
trajectory, prediction, 241 notetaking/clinical documentation,
use cases, 214–215 GenAI (usage), 148
Generative design, 9, 178–184 operational improvements,
enhancement, GenAI 143, 147–149
(usage), 179–181 outbreaks, predictions, 149
GenAI generation, 12–13 professionals, job loss
Generative FAQ for CCAI Insights (potential), 70–71
(Google Cloud), 111 routine calls/messages (answering),
Generative models, examples, 7–8 conversational AI (usage), 148
Generative music, 157 transformation, 143
Generator-enhanced optimization Health research, Google DeepMind
(GEO) model, BMW usage, 30 support, 178
Gene technology, GenAI (impact), 25 HeyGen, 256
Genetically modified organizations Hopla advice robot (Carrefour), 127
(GMOs), GenAI HSBC, GenAI usage, 195
(impact), 25–26 Hugging Face, 262
Genny, (voice-generation tool), 88 Human creativity, stimulation, 5
GenRocket, 266 Human customer service agents,
Ghostwriter GenAI tool (Ubisoft), 158 support, 110
Github Copilot GenAI, 262 Human + Machine: Reimagining Work
copyright infringement, 54 in the Age of Ai (Daugherty/
usage, 205–206, 208 Wilson), 111
GoArt (AI image generation tool), 92 Human resources (HR), GenAI
Gradescope (education tool), 253 (usage), 37–38

277
 Index

Hybrid models, emergence, 22 risk, 63–69

Hyro (HIPAA-compliant lower-level jobs, higher-level jobs
conversational AI), 148 (differences), 64–66
Journalism, GenAI (relationship), 79–81
I
Imagebird (Meta), 250 K
Images, GenAI generation, 9, 11 K2View, 267
Inbreeding, 47 KeyBank, GenAI usage, 195
Influencers, transparency Khan Academy, OpenAI
issue, 104–105 partnership, 136
Inner Life of an AI, The: A Memoir Knowji (education tool), 253
by ChatGPT (GenAI Knowledge graph, access, 40
creation), 83–84 Knowledge work (white collar work),
Insilico Medicine, GenAI (usage), 176 GenAI augmentation, 71
Instance-Conditioned Generative Know Your Customer (KYC), GenAI
Adversarial Networks” assistance, 192
(IC-GAN), Meta usage, 37 KopiKat, 267
Interactive problem-solving, GenAI Kortext (education platform), 135
(usage), 116 Kyoto Date, creation, 104
Internal accounting functions, GenAI
assistance, 192–193 L
Internet of Things (IoT), 24 LaLiga Tech, Globant/Microsoft
Intuit Assist (QuickBooks), AI helper (GenAI partnership), 83
(introduction), 197 Language Model for Dialogue
Invideo, 256 Applications (LaMDA)
Ivy (education tool), 253 (Google), impact, 21–22
Large language models
J (LLMs), 7, 21, 40
Jasper (content creation tool), 38, 251 Copilot (Microsoft), 218
JetBlue, ASAPP partnership, 112 Harvey construction, 168–169
Jewelry/fashion, AI examples, 182–183 Redfin usage, 206
Jobs Law firms, GenAI (usage), 166–169
AI augmentation, 71–72 Lawyers, job loss (potential), 70–71
automation risk, 66–67 Learners
creation, GenAI (impact), 73–74 learning assistance, 135–137
GenAI, impact, 63 learning process
learning ability, loss (potential), 69 (rethinking), 133–137
loss Legal advice chatbots, GenAI
moderate/low risk, 69–72 (usage), 166

278
 Index

Legal cases, predictive analysis materials (creation), GenAI

(usage), 166 (usage), 99–102
Legal contract analysis, AI “Masterpiece” (Coca-Cola ad), 100
(usage), 166 Materials science, GenAI (impact), 25
Legal documents McAfee, Andrew, 227
AI-assisted creation/review, 165 MDClone, 267
understanding, GenAI Media
usage, 169–170 GenAI examples, 80–81
Legal profession, GenAI transformation, 79
examples, 167–169 Medicine
Legal research, GenAI (usage), 166 GenAI examples, 177–178
LegalRobot, usage, 169 personalized medicine, enabling, 176
Lemoine, Blake, 21 Merchat AI (Mercari), 128
Lex (AI co-pilot), 251 Messi, Lionel (PepsiCo deal), 103
Lil Miquela (virtual influencer), 104 Metabob, 262
Lisa (Mango GenAI platform), 130 Metaverse
Lopez, Javi, 159 GenAI, relationship, 24–25
Loudly (royalty-free music immersive shopping, 122
generator), 88, 259 MicroStrategy AI (GenAI
Lovo.ai (voice-generation tool), 88 product), 219, 264
Lower-level jobs, higher-level jobs Midjourney (GenAI tool), 11, 38,
(differences), 64–66 92, 159, 255
Loyalty programs, tailoring, 128 Mintlify, 263
Lu do Magalu (virtual influencer), 104 Misinformation, potential, 46–47
Moody’s, Google Cloud
M partnership, 196
Macfarlanes, Harvey Morgan Stanley, GPT-4 usage, 195
partnership, 168–169 Mostly, 267
Machine learning, 6–7, 17 MSN, human journalists (GenAI
Machine vision, 20 replacement), 81
MagicSchool (GenAI tool), 135 Mubert (AI music generator), 88, 259
Mann, Scott, 87 Murf, 259
Marketing Muse generative platform, 160
AI, usage, 97 MuseNet (AI music generator), 88
content creation, Music
streamlining, 38–39 AI-driven consumption, 89–90
GenAI AI generators, 88
advantages, 98 GenAI generation, 9, 12, 87–90
application, 97–99 generative music, 157

279
 Index

MusicLM, 260 authors, legal action, 54

Mutable, 263 GPT models, 21
My Assistant (Walmart), usage, 36–37 Operational risk management, GenAI
assistance, 192
N OpusClip, 257
Narrow AI (applied AI), 241 Organization
Natural language data analysis, GenAI (usage), 217
processing, 20 executive roles, creation, 229–230
summaries, GenAI (impact), 13 GenAI, deployment, 130
NatWest, IBM collaboration, 194 Organizational culture/structure,
Netflix, thumbnails, 102–103 GenAI (relationship), 228–229
NetSuite (Oracle), enterprise resource Original thought, ChatGPT/DALL-E
planning (ERP) applications, 197 capability (absence), 64
Networks, 235–236 Outfit Creation Model (OCM), Stitch
Neural networks, 6–7, 20–21 Fix usage, 32
layer organization, 7
learning, data exposure, 6 P
output generation, 7 Pahdo Labs, GenAI tool launch, 162
training, 6 Pathways Language Model (PaLM)
Newegg, 123, 126–127 (Google), 22
News Corp, GenAI (usage), 81 Patient-doctor conversations,
Newsreader (creation), GenAI supplementation, 146
(usage), 65 Patients
Next-generation internet personalized advice, 143, 144
searching, 39–40 personalized/improved treatment,
Ninja Theory, GenAI usage, 158 providing, 146–147
Non-fungible tokens (NFTs), self-examination, assistance, 145
auctions, 91–92 Pattern creation, GenAI (impact), 5
Non-player characters (NPCs) Payments industry, enhancement
conversational skills, developer (GenAI usage), 196
improvement, 161 PC Builder (Newegg), 127
platforms, 160 Peech, 257
usage, 155, 156 Peretti, Jonah, 80
Noonooouri, 89, 103 Personalization, 102–103, 112–114
Notta, 269 GenAI, impact, 190
Personalized experiences,
O examples, 102–103
Octopus Energy, ChatGPT Personalized fan engagement, PGA
usage, 39, 109 (connection), 113–114
OpenAI Personalized learning, 133

280
 Index

Personalized medicine, enabling, 176 Progressive GANs (NVIDIA),

Personalized treatment, 143 development, 21
Personalized user experiences, GenAI Project Gutenberg, Microsoft
(usage), 116 collaboration, 85
PGA Tour, personalized Promotions/discounts, tailoring, 128
fan engagement Proprietary data, importance, 234–235
(connection), 113–114 Prosecraft, shutdown, 54
Physical stores, Publishing process (enhancement),
augmentation, 124–125 GenAI (impact), 84
Pictory, 257
Plaito (education tool), 253 Q
Podcasts, GenAI (usage), 83, 85 Qlik Cloud platform, ChatGPT
Powell, Kimberly, 174–175 experience, 218, 265
Power BI (Microsoft), 264 Quantitative trading, GenAI
Power Platform (Microsoft), 112 usage, 191
Predictive AI construction, Dynatrace Querium (education tool), 253
efforts, 203 Quizlet (GenAI study aid), 136
Predictive fintech solutions, 115
Predictive maintenance, GenAI R
(usage), 116 Random data (noise), addition, 8
Privacy preservation, GenAI Randomness, GenAI (impact), 7
(impact), 14 Raymond, Jade, 158
Proactive interventions (delivery), Real-time data, importance, 233–234
predictive AI (usage), Real-time reporting, GenAI
114–115 (impact), 13
Products Recursion Pharmaceuticals, NVIDIA
creation, embedded GenAI collaboration, 175–176
(usage), 33–35 Repilt, 263
descriptions, generation, 129–130 Research, GenAI generation, 9, 15
design Retail
GenAI, impact, 13 reimagining, 121
personalization/uniqueness, 126 uses, 129–130
GenAI Revoicer (GenAI text-to-speech
incorporation, 1–11815 platform), 53
infusion, 35–36 Revolution Software, success, 159
innovation, 33–36 RhythmX AI platform, 146–147
personalization, 36 Richardson, James, 212
Programming Robotics
AI, impact, 201 GenAI, impact, 25
GenAI, impact, 202–205 NVIDIA research, 244

281
 Index

Runway, 257 Societies, transformation, 29–31

Russell Group universities, GenAI Software developers, future need
guiding principles, 139 (question), 203–204
Rytr, 251 Sogeti, 267
Sound
S design, GenAI (impact), 12
Salesforce Anypoint Code Builder effects, GenAI (usage), 157
(coding tool), 208 Soundful (AI music
Saman Bank, GenAI usage, 194 generator), 88, 260
SAP Build Code (coding tool), 208 Sound, GenAI generation, 9, 12
Scenario analysis, GenAI (usage), 191 Soundraw, 260
Schools, formal policy/guidance Special effects, GenAI (impact), 11
(need), 138–140 Speechify, 260
Scribe, 252 Splash, 260
Search Generative Experience (SGE) Sports
(Google), 39 broadcasting (enhancement),
Sectors, transformation, 29–31 GenAI (usage), 81–83
Security Co-Pilot (Microsoft), 270 GenAI application, 81–82
Sentis generative platform, 160 GenAI examples, 82–83
Services SQL-GPT (Kinetica), 219
creation, embedded GenAI Square (Block Inc.), GenAI
(usage), 33–36 usage, 190, 194
GenAI, infusion, 35–36 Stability AI, copyright law
innovation, 33–36 breach, 53–54
personalization, 36 Stable Diffusion (GenAI tool), 11,
ShopBot (eBay), 127 22, 92, 255
Shopping journeys, Storytelling, GenAI (author
personalization, 126–127 usage), 83–84
SHRDLU (AI program), 18 Storytelling, GenAI (usage),
Shudu Gram (avatar supermodel), 83–85, 157
Fenty collaboration, 104 Stripe, GPT-4 usage, 196
Simplified Molecular Input Line Entry Students, value (GenAI
System (SMILES), 174 addition), 134–135
Simulation/modeling, GenAI Studio Bot, 263
(impact), 14 Suleyman, Mustafa, 242
Smart mirrors/fitting rooms, Supplier negotiations, automation,
usage, 125 130
Snapchat, My AI bot usage, 40–41 SyncDreamer GenAI tool, 160
Social scoring, 56 Synthesia, 257

282
 Index

Synthesys (voice-generation U
tool), 89, 258 Udemy (online learning), 232
Synthetic data Uizard (coding tool), 208
GenAI generation, 9, 14–15
usage, 215 V
Synthetic voice, GenAI (impact), 12 Variational autoencoders, usage, 7, 20
Syntho, 268 VEED:IO, 258
Vertex AI Vision System (Google
T Vertex), 249
Tableau, GenAI usage, 218, 265 sports broadcaster usage, 83
Tableau Pulse, 218–219 Video games
Tabnine (coding tool), 208, 263 design/testing, GenAI
Teachers approach, 155–161
job loss, potential, 70–71 developers, GenAI tools, 160–161
teaching development
assistance, 135 elements, 156–157
process, rethinking, 133–137 GenAI uses, 156–158
Teaching content, future, GenAI (impact), 161–162
transformation, 137–138 GenAI examples, 158–159
Technology, usage/ GenAI generation, 9, 15
placement, 235–237 Video, GenAI generation, 9, 11–12
TELUS, HEAVY.AI partnership, 217 Virtual AI shopping
Terradata, GenAI natural language assistances, 127–128
interface, 218 Virtual reality (VR), GenAI
Text, GenAI generation, 9, 10–11 generation, 15
Thomson Reuters, GenAI usage, 169 Virtual storefronts, 122
Tonic, 268 Virtual try-ons, 121, 123–124
Traditional AI Virtual Volunteer (chat/image-to-text
generative AI, contrast, 4 recognition tool), 34–35
learning/output, 5 Virtual worlds, GenAI generation, 9
Transformer models, 20 Visual representation, GenAI
Trip/itinerary planning, GenAI (impact), 13
(usage), 42 Voice
Tripnotes, usage, 42 AI generators, 88–89
TrueSync (Flawless), GenAI Altered AI voice library, usage, 158
(usage), 87 modification, GenAI (impact), 157
Turing, Alan (Turing Test), 18 re-creations, perception
Tynker Copilot (coding tool), 208 problems, 86
Type Studio, 258 synthesis, GenAI (impact), 12, 157

283
 Index

W What the Diff (coding tool), 208

Walmart Wilson, H. James, 111
employee, time (saving), 36–37 Wordtune, 252
GenAI shopping assistant, WriteSonic, 252
127–128
Warp, 264 X
Watermark, usage, 48 Xiao, Forrest, 84
WatsonX Code Assistant (coding
tool), 208, 262 Y
WaveNet (Deepmind), YData, 268
development, 20–21
Wavtool, 261 Z
Website experience, ZeroGPT, 48
improvement, 122–123 Zuber, Joerg, 89

284
 WILEY END USER LICENSE
AGREEMENT
Go to www.wiley.com/go/eula to access Wiley’s ebook
EULA.