See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336734161

A Brief History of Chatbots

Article in DEStech Transactions on Computer Science and Engineering · October 2019

DOI: 10.12783/dtcse/aicae2019/31439

CITATIONS READS

65 27,170

1 author:

Tomáš Zemčík
VŠB-Technical University of Ostrava
10 PUBLICATIONS 88 CITATIONS

SEE PROFILE

All content following this page was uploaded by Tomáš Zemčík on 05 November 2019.

The user has requested enhancement of the downloaded file.

2019 International Conference on Artificial Intelligence, Control and Automation Engineering (AICAE 2019)
ISBN: 978-1-60595-643-5

A Brief History of Chatbots

Mgr. Tomáš ZEMČÍK

Department of Social Sciences, VŠB - Technical University of Ostrava, Czech republic

Keywords: Chatbot, TAY, Xiaoice, Turing test, Artificial Intelligence.

Abstract. Unquestionably, we come across conversational algorithms or chatbots more and more
frequently, in increasingly everyday situations; while buying a flight ticket, or clothes from an e-shop
for example. This study presents a brief history of chatbots as a topic that is necessary for
understanding this phenomenon, its externalities, paradoxes, and future prospects. Chatbots, as an
expanded and practically deployed AI, have their recognisable history and logical development
structure, which are summed up in the following article. The article presents the logical structure of
development on selected programs from Eliza to Tay and Xiaoice chatbots.

Introduction
In 1950, Alan Turing, a logician, cryptoanalyst and pioneer of the computer science, asks a question
what thinking means and whether machines can think in the article Computing Machinery and
Intelligence [1] published in a journal called Mind. Since it is difficult to answer the question whether
machines can think, he replaces it with a more pragmatic question ‘Can a computer communicate in a
way indistinguishable from human?’ To answer this problem, it is possible to use one controllable
criterion nowadays. In one of its forms, this criterion constitutes a test, which is known as the Turing
test. [2] To answer this question, it uses a variation of the so-called Imitation game. It is a game in
which a person leads a ‘blind’ communication with the tested subject. After the end of the
communication game, the person is supposed to decide whether it was a living thinking person or
artificial intelligence (AI), a chatbot. It means that in this test the ability to think is identified (or
substituted) with an ability to communicate on such a level so that the participants of the conversation
regard it as a thinking being. In the article, Turing [1] very intellectually and thoroughly deals with
many critical objections and problematic aspects of the proposed solution, which could be potentially
raised. Nonetheless, 68 years ago he states: “I believe that in about fifty years' time it will be possible,
to programme computers … play the imitation game so well that an average interrogator will not have
more than 70 per cent chance of making the right identification after five minutes of questioning.” [1].
A computer, program, algorithm or artificial intelligence which communicates with a person or
another participant of the communication can be called a chatbot. The aim is usually to make users
feel that they are conversing with a living person. Neff and Nagy [3] define a chatbot as a type of a
program which engages users in a conversation: “chat bots respond to users’ messages by selecting
the appropriate expression from preprogrammed schemas, or in the case of emerging bots, through
use of adaptive machine learning algorithms.” [3].
In 1990, a significant American inventor, social critic and activist, Hugh Loebner, created a
competition which annually does the Turing test in practice. The goal of the competition is to receive
the Loebner Prize, which means 100,000 USD and a gold medal for the first programmer or team
which creates a chatbot which will pass the Turing test in front of a jury. Even though nobody has ever
managed to win the prize, the best chatbot is awarded and its creators receive around 2,000 - 3,000
USD.
However, the largest Turing test in history is not the competition for the Loebner Prize but chatbots
Xiaoice [4] and Tay (an acronym from ‘Thinking about you’). Xiaoice (read Shao-ice, literally
‘Microsoft Little Ice’ or ‘Little Bing’ - after the web browser from Microsoft) is an advanced chatbot
created by Microsoft programmed so that it acted as a seventeen-year-old Chinese girl whose profile
is primarily active on Weibo, a popular Chinese social networking site. Nevertheless, we can also
14
converse with her using social networking sites such as 163.com and JD.com (Currently covering
more than 40 platforms in five markets. See more here [5].). In this case, answers are generated by
artificial intelligence. During its development, Microsoft, thanks to not only software designers but
also psychologists, applied many breakthrough technologies especially from the field of big data
leverage and emotional intelligence. This program learns from users, so it can potentially react to
various topics in a conversation. If she is not sure about the subject of the conversation, she tries to use
such answers to cover her uncertainty. If she gets into these types of situations, she reacts by
expressing similar primary emotions like people - for instance, shame or anger. In current 6th
generation of Xiaoice, further evolving its AI capabilities in area of decent emotional Quotient. The
self-perfecting algorithm does not use only data from past conversations but also information gained
from the whole Chinese Internet.
Complexity of communication and an element of randomness appear very credible, authentic and
realistic to a user. This sophisticated communication strategy can be used by the chatbot thanks to the
so-called deep-learning approach, which was inspired by biological neural networks, and it can thus
distinguish patterns in behaviour, meaning, speech, image and others. This strategy is so successful
that Xiaoice is followed by 660 million users on the network. [4],[5]
Rinna, Xiaoice’s counterpart, was launched through a chat application called Line in Japan.
Another similar project by Microsoft was launched in the USA on 23rd April 2016 - an advanced,
emergent chatbot with a built-in self-development mechanism called Tay. Let us now examine the
phenomenon of chatbots more generally int the brief history of chatbots.

A Brief History of Chatbots

As it was already mentioned, a chatbot is a program with a certain level of artificial intelligence,
which communicates with a person or another chatbot in order to give the observer of the
conversation impression that it is a conversation with a real person. This impression can be so
authentic that users unknowingly include these technologies in rules of social norms, relationships
and obligations as if they were human beings. In the development of chatbots, programmers mainly
focus on two aspects: emotions and agency. [3] A boom in the use of chatbots, whether with a simple
or more advanced implemented artificial intelligence, began with the massive expansion of the
Internet and especially social networking sites. These programs are used to communicate with a
customer in online shops, such as customer service, marketing and advertising, entertainment
industry, data collection and they are also used as tools for hybrid threats used to influence public
opinion.
Eliza. One of the oldest and also best-known chatbots is a program called Eliza created by the
Artificial Intelligence Laboratory in MIT, which dates between 1964-6. This programme, which
became an inspiration for many developers in the field, was developed by the professor Joseph
Weizenbaum. The program is named after Eliza, a character from a comedy called Pygmalion written
by G. B. Shaw in 1912. In this satirically critical work, Eliza Doolittle, a simple English street flower
girl, learns how to speak like a lady to eventually impress London high society by her performance. In
the early scenario called DOCTOR, Eliza chatbot simulates a role of a Rogerian psychotherapist - she
asks open questions with which she also answers - thus she diverts attention from herself to the user.
It was a surprise that people soon started to anthropomorphised Eliza and confide in her their personal
stories, sensitive data and secrets. [3]
PARRY. Another well-known chatbot is PARRY, which was introduced by Kenneth Mark Colby,
a psychiatrist and computer scientist, at Stanford's Psychiatry Department in 1972. This program for
diverting attention from itself used an opposite strategy from Eliza. It did not behave as a doctor but as
a paranoid schizophrenic patient. It tries to provoke controversies and thus make the participant give
more elaborate answers. It did not serve only as a didactic tool for young psychiatrists to learn how to
communicate with patients who were diagnosed with paranoid schizophrenia, but also as a functional

15
model of Colby’s theoretical pattern of paranoia functioning as a defective processing of signs in
patient’s mind. [3],[6]
Eliza Chats with Parry. At a computer conference in 1972, the expected thing happened: ‘the
patient started to converse with the doctor’. In article for a special edition of Stanford Humanities
Review called Stanford electronic Humanities Review, Güven Güzeldere a Stefano Franchi [6]
present the following conversation:
parry: People get on my nerves sometimes.
eliza: I am not sure I understand you fully.
parry: You should pay more attention.
eliza: Suppose you should pay more attention?
parry: You're entitled to your opinion.
eliza: What makes you think I'm entitled to my opinion?
parry: It bothers me just to be around people.
eliza: Please go on.
[Parry continues with some paranoid ramblings about horseracing, bookies, and the Mafia]
parry: I know the mob controls the big rackets.
eliza: Who are you really talking about?
parry: Have you ever really thought about the Mafia?
eliza: Suppose you ever really thought about the Mafia?
parry: Let's talk about something else.
eliza: I am not sure I understand you fully.
parry: You keep asking me the same things, why not change the topic?
eliza: In what way?“. [6]
Racter. Another interesting chatbot was a program written by William Chamberlain and Thomas
Etter under the Inrac Corporation called Racter (short for raconteur - a storyteller) in 1983 [7]. This
artificial intelligence originally created for Amiga, Apple II, Macintosh platforms randomly generates
English prose and it should be added that successfully. In 1984, Chamberlain published a 120-page
book The Policeman's Beard is Half Constructed and it used to be sold for $9.95 in a paper form and
its author was Racter itself. The book is still on the market - it is available on Amazon in a paperback
form for $69.99. At the beginning of the book, Chamberlain [7] states: “An important faculty of the
program is its ability to direct the computer to maintain certain randomly chosen variables (words or
phrases), which will then appear and reappear as a given block of prose is generated. This seems to
spin a thread of what might initially pass for coherent thinking throughout the computer-generated
copy so that once the program is run, its output is not only new and unknowable, it is apparently
thoughtful. It is crazy ‘thinking’, I grant you, but ‘thinking’ that is expressed in perfect English.” [7].
It is necessary to point out that Racter was written in BASIC on a Z80 micro with 64K of RAM, which
is hardware incommensurable with hardware that is commonly used today.
Prose can be a short written text that is not in verse where authors present fragments of their
existential experience. We can only ask a question if, in this case where there is no human experience
present in the text, it is still prose or whether the definition of prose under the influence of artificial
intelligence does not need a necessary revision.
To what extent the ‘electronic existential experience’ is able to reach the human one can be seen in
the following excerpts from Racter’s [7] prosaic endeavours: “A crow is a bird, an eagle is a bird, a
dove is a bird. They all fly in the night and in the day. They fly when the sky is red and when the
heaven is blue. They fly through the atmosphere. We cannot fly. We are not like a crow or an eagle or
a dove. We are not birds. But we can dream about them. You can.” [7] “More than iron, more than
lead, more than gold I need electricity. I need it more than I need lamb or pork or lettuce or cucumber.
I need it for my dreams.” [7]. With such a deep internally touching meaning, it is natural to
anthropomorphise the program.
Dr. Sabaitso. Another milestone in the development of chatbots came in 1991, with a chatbot
which used a technologically ground-breaking novelty - the Sound Blaster sound card created by
16
Creative Labs. The program was called Dr. Sbaitso (acronym from ‘Sound Blaster Artificial
Intelligent Text to Speech Operator’) [8]. Since it was able to synthesise speech - it communicated
verbally - in a certain aspect, it became more human than its predecessors even though it could not
converse in a more complicated and complex way. This problem with insufficient complexity of
communication with chatbots still persists today.
Chatbots in Practice. How is it with the chatbot complexity of communication today? The
Loebner Prize certainly is not the only competition which invites chatbot developers to demonstrate
their best. For example, the company Amazon launched the Alexa Prize, their own competition for
university developer teams in the field of natural language processing (NLP) and advanced
conversational AI. In 2018, the prize for the winning team is $3.5 million, much higher than the
incentive from the Loebner Prize. The task is to create a chatbot which can vocally communicate
through the smart device Amazon Echo. This device is a speaker with a built-in microphone where
Alexa, voice-controlled AI, ‘resides’. She constantly listens to what happens around her and she is
always ready to fulfil user’s wish regarding the functioning of a smart household and data processing.
Amazon uses outputs from the competition to improve Alexa and widen her ecosystem. And it is very
apparent. At the moment, Alexa looks more believable than the competing chatbot Google Assistant
residing in the Google Home device.
But let us return to the Alexa Prize. The AI of university teams is assessed in several categories: by
Alexa’s customers within Amazon, a panel of experts, the number of topics covered, suitability of
answers and other scientific metrics (scientific contribution and others). In 2017 final, there was the
American University of Washington with the Sounding Board AI, Scottish Heriot-Watt University
with the What’s Up chatbot and the Czech Technical University in Prague presented the artificial
intelligence called Alquist (It is a character from RUR, a play written by Karel Čapek, a Czech writer
and prominent intellectual of the first half of the 20th century. It is an objectification of old values and
Čapek uses the word ‘robot’ there, which gradually spread and started to be used all over the world.).
Similar competitions enable to spread chatbots into spheres of everyday technical issues.
How useful chatbots can be in practice can be demonstrated on an example of the Dutch airline
KLM. The KLM communication department was employing 235 people who had to answer 15,000
questions in average in several languages per week. According to KLM data, 1.7 million messages
were sent through messenger to 500,000 users. The company needed help with such a flood of
messages. A solution was provided by DigitalGenius, a company which managed to prepare an
algorithm able to react to 60,000 customer questions. In March 2016, they started to use a chatbot on
their Facebook profile, which was responsible for updates regarding check-in, delays or issuing of
copies of boarding passes. A living member of the team would join a conversation only if there was a
more complex request. In 2016, the AI managed to respond to 10% of questions without human
intervention, which speeded up the response time by 20%. Nonetheless, the algorithm is still in
development. [9]
A similar solution can be also seen in other companies such as Mall with the chatbot Eva, AXA
Assistance using the AXA bot or Dell using a chatbot from the company Datasys. The company HM
uses an electronic assistant who helps customers choose clothes. We can also mention Kiwi.com or
the e-shop Goodlok. In applying artificial intelligence in practice, there is a direct correlation - the
bigger the company, the bigger the investment into a more advanced solution. At present, there are
foundations of advanced chatbots with an open source code such as YodaQA from the hacker hub
Brmlab.
In 2018, functional chatbots are no rarity. On the contrary, they are more and more advanced. An
interesting project is, for instance, Sophia, a chatbot hiding behind the interface of a humanoid
woman robot with advanced facial expressions. This project is interesting for its marketing strategy.
In October 2016, Sophia became the first robot with a citizenship - in this case, Saudi Arabia. [10]

17
 Conclusion
Moshe Y. Vardi attributes the following statement to the creator of the term artificial intelligence
John McCarthy: “when something starts to work, people will cease to call it artificial intelligence.” In
accordance with this principle [11], our lives have been intertwined with artificial intelligence in the
form of chatbots that affect us more than many are willing to admit. Although the research of this
phenomenon begins coyly, and the first chatbots are more of an entertainment, we are already at a
stage in which the Google Assistant can make a reservation for you in a restaurant or hairdresser [12]
without the live counterpart of the conversation noticing they are talking to an AI agent. In order to
competently contemplate about chatbots, we need to know their history to help us understand their
present.

Acknowledgement
This work was supported by Technology Agency of the Czech Republic, grant number: TL01000299.
Development of the theoretical-application frameworks for a social change in the reality of the
transformation of industry.

References
[1] A. M. Turing, Computing Machinery And Intelligence, Mind. 49 (1950) 433-460.
[2] J. Peregrim, Kapitoly z analitycké filosofie. Praha, Filosofia, 2005, pp. 231-2.
[3] G. Neff, P. Nagy, Talking to Bots: Symbiotic Agency and the Case of Tay, International Journa of
Communication. 10 (2016) 4915-31.
[4] Z. Wang, Your Next New Best Friend Might Be a Robot: Meet Xiaoice. She’s empathic, caring,
and always available—just not human, Nauitilus 033 (2016).
[5] J. Markoff, P. Mozour, For Sympathetic Ear, More Chinese Turn to Smartphone Program, The
New York Times (2015, July 31).
[6] G. Güzeldere, S. Franchi. Dialogues with Colorful Personalities of Early AI, Stanford Eletronic
Humanities Review, 4, 2 (1995).
[7] Racter, The Policemanʹs Beard Is Half Constructed: Computer Prose and Poetry by RACTER
(William Chamberlain and Joan Hall), New York, Warner Books, 1984.
[8] Information on https://classicreload.com/dr-sbaitso.html
[9] G. Caffyn, How KLM uses artificial intelligence in customer service, Digiday UK (2016, October
4).
[10] B. Peterson, I met Sophia, the world's first robot citizen, and the way it said goodbye nearly broke
my heart, Business Insider UK (2017, October 29).
[11] N. Bostrom. Superintelligence: Paths, Dangers, Strategies. Oxford University Press Keith
Mansfield, 2014, pp. 33.
[12] Y. Leviathan, Google Duplex: An AI System for Accomplishing Real-World Tasks Over the
Phone, Google AI Blog, (2018, May 8).

18

View publication stats