Artificial

Intelligence
Harvir
Singh
12301306
 Contents

Notion of intelligence, levels and types of artificial

Introduction:
intelligence, importance,
advantages, and limitations of artificial intelligence,
uses and applications of artificial intelligence.

Intelligent structure of intelligent agent, agent performance,

rational agent, types of agent
Agents: environment, agent architectures.

Problem solving State space search, informed search strategies,

uninformed search
using Search: strategies, production system characteristics.
 01
Introduction
 Notion of
Intelligence
Intelligence is about learning new things and
using that knowledge to solve problems. When
we talk about machine intelligence, it means a
machine can do things like a human—learning
from past experiences, recognizing patterns, and
making decisions using the information it has.
Type ● Artificial Narrow Intelligence (ANI): Also called
weak AI, ANI is built to do just one thing or a small

s of set of things, like how virtual assistants like Siri or

Alexa can help with specific tasks.

AI ● Artificial General Intelligence (AGI): AGI, or strong

AI, refers to machines that can think, learn, and
apply knowledge across many different tasks, just
like a human can.

● Artificial Superintelligence (ASI): ASI is when a

machine becomes smarter than humans in every
way, including creativity, wisdom, and social skills.
 Importance of AI
• In Medical Science: Artificial Intelligence has revolutionized the
medical field, transforming how we diagnose and treat patients. AI can
predict crucial outcomes, like whether a patient has malignant or benign
cancer, using symptoms, health records, and history. It's also being used
to warn patients about potential health declines and guide them on how to
get back to a healthy life.

• In Air Transport: Air transport, one of the world's most complex

systems, now relies on AI to optimize operations. AI helps plan flight
routes, manage landing schedules, and ensure takeoffs run smoothly,
making the process more efficient.

• In Banking and Financial Institution: AI is crucial in managing

financial transactions and other banking operations. From handling daily
transactions to managing stock markets and financial operations, AI makes
these tasks more efficient and reliable.

• In Gaming and Entertainment:

AI has driven huge advancements in
gaming, from virtual reality experiences to modern video games. AI-
powered bots are always ready to play, so you never need to wait for
another person to join in.
 • Efficiency
Advantages of AI
Boost: AI can handle repetitive tasks faster and more
accurately than humans, freeing us up for more creative work. Think of it
as having a tireless assistant that never needs a break!

• 24/7 Availability: Unlike humans, AI doesn’t need sleep. It can operate

around the clock, ensuring that critical processes continue without
interruption.

• Data-Driven Decisions: AI can analyze vast amounts of data quickly,

spotting patterns and trends that might be invisible to us. This leads to
smarter, more informed decisions.

• Personalization: From recommending the perfect movie to customizing

your shopping experience, AI tailors services to your individual
preferences, making everything feel just right.

• Innovation Catalyst: AI powers breakthroughs in fields like

healthcare, where it helps in early disease detection and personalized
treatment plans, pushing the boundaries of what's possible.
•
Limitations of AI
Lack of Creativity: While AI can process data and make decisions,
it doesn’t have human creativity. It follows rules and patterns but
can’t think outside the box.

• Dependence on Data: AI is only as good as the data it’s trained on.

If the data is biased or flawed, the AI’s decisions will be too.

• High Cost: Developing and implementing AI can be expensive. It

requires significant resources, both in terms of time and money.

• Job Displacement: As AI takes over more tasks, there’s a growing

concern about job loss in certain sectors, especially those
involving repetitive or routine work.

• Ethical Concerns: AI raises questions about privacy, decision-

making transparency, and the potential for misuse. As AI becomes
more powerful, these concerns become more pressing.
 early detection,
Healthcare personalized care,
Revolution robotic surgery

Uses and Smart Home

Assistants
smart devices,
daily tasks, voice
control
Applicatio Manufacturing &
production
optimization,

ns of AI
predictive
Industry maintenance,
quality contro
chatbots, virtual
Customer Support assistants, 24/7
support
Intelligent
Agents
 What are intelligent agents?

An intelligent agent These agents are Think of intelligent

is like a smart robot key to AI because agents as clever
that can sense what's they interact with helpers that use
happening around it, their environment their senses to
think about what it much like humans figure out the world,
sees, and then take do, making make decisions, and
action to reach its decisions, learning act on them—just like
goals. from experiences, how we navigate
and solving through our day and
problems. tackle challenges.
 Agent Performance
We judge how well an agent is doing by looking at how effectively it
reaches its goals in its environment. To evaluate its performance, we
check how accurately it acts, how efficiently it uses resources, and
how quickly it makes decisions.

The best agents excel by striking the perfect balance between

accuracy, efficiency, and speed, making them the superheroes of
problem-solving!

They adapt to new situations with ease and continuously improve their
strategies to tackle even the toughest challenges.
Rational
Agents
A rational agent is like a
smart strategist, always aiming
to get the best possible
results. When things are
uncertain, it strives for the
most promising outcome. We
measure its smarts by looking
at how well it uses its
knowledge, the choices it
makes, and how likely those
choices are to succeed. The
best agents don’t just react;
they plan their moves
carefully, adapt to new
information, and continually
refine their approach to stay
 TYPES OF AGENT ENVIRONMENT

Fully Observable vs Deterministic vs Deterministic vs

Partially Stochastic : In a Stochastic : In a
Observable : In a fully deterministic deterministic
observable environment, the next environment, the next
environment, the state of the state of the
agent has access to environment is fully environment is fully
complete information determined by the determined by the
about the current state and the current state and the
environment. In a agent's actions. In a agent's actions. In a
partially observable stochastic stochastic
environment, the environment, there is environment, there is
agent only has some level of some level of
incomplete or noisy randomness in the randomness in the
 Types of agents

Goal-Based Agents: Learning Agents :

These agents take actions that help achieve a predefined goal. These agents can learn from their experiences and improve their
They decide on their actions based on the goal they need to performance over time. They have components like a
accomplish, not just the current state of the environment. learning element, performance element, critic, and problem
Example: A navigation system that plans a route to a generator. Example: A machine learning model that
destination. improves its accuracy as it gets more data over time.

Utility-Based Agents: Multi-Agent Systems :

These agents not only aim to achieve goals but also consider These involve multiple agents that can interact with each other,
which goal has the highest utility or "happiness." They use either cooperatively or competitively. Each agent may be
a utility function to evaluate different outcomes and choose simple, but the overall system can solve complex problems.
the action that maximizes utility. Example: An AI for Example: Autonomous drones working together to
financial trading that not only seeks profit but also tries to complete a search-and-rescue mission.
minimize risk.
 Agent Architecture
Hybrid
Architecture :
Hybrid architectures combine two or more of the above architectures to leverage the strengths of each. For
example, an agent might use a reactive system for quick responses and a goal-based system for long-
term planning. Example: An autonomous vehicle that uses reactive architecture for real-time obstacle
avoidance and goal-based architecture for route planning.
Learning
Architecture:
Learning agents have the ability to improve their performance over time based on experience. They
incorporate machine learning algorithms to adapt their behavior based on feedback from the environment.
Example: A recommendation system that personalizes content based on user interactions.

Reactive (or Simple Reflex)

Architecture:
Reactive architectures consist of condition-action rules (also known as production rules). They do not
maintain an internal state and simply respond to stimuli from the environment with pre-programmed
responses. Example: A basic robotic controller that avoids obstacles based on sensor input.
 Problem
solving using
search
 State Space Search
State space search involves exploring possible states to find
a solution to a problem. It starts with an initial state and
applies operators to generate new states until a goal state is
reached.

Components of State Space Search:

 State: A representation of a particular configuration of the

problem at a specific point in time.
 Initial State: The state where the search begins.
 Goal State(s): The state(s) that satisfy the problem's goal
condition.
 Actions: The set of operations that can be performed to
transition from one state to another.
 Transition Model: A description of how actions change
the state.
 Path Cost: The cost associated with moving from one
state to another via a specific action. The cumulative path
cost is often used to find the most efficient solution.
 Informed Search Strategies:
■ Informed search strategies use heuristic functions to guide
the search process, helping to find solutions more
efficiently.
■ A* Search : Uses a heuristic to estimate the cost from the
current state to the goal, optimizing the path cost.

■ Greedy Best-First Search : Selects the node that appears

to be closest to the goal based on a heuristic.
■ Heuristics in Informed Search :A heuristic is admissible
if it never overestimates the true cost to reach the goal.
■ Hill Climbing:A search strategy that continuously moves
to the neighbor with the highest heuristic value.
■ Beam Search:A search strategy that explores a fixed
number of the best nodes at each level of the search tree
(beam width) rather than all nodes.
 PRODUCTION SYSTEM
CHARACTERISTICS

A production system consists of a set of rules, a database and a control

system. It operates by applying rules to the database to generate new
knowledge or solve problems.

The main characteristics include:

– Simplicity : The rules are usually straightforward and modular.
– Flexibility : New rules can be added to enhance the system's
capabilities.
– Transparency : The reasoning process is often easy to follow, as it
is based on explicit rules.
THANKS
Harvir
Singh
12301306