This Photo by Unknown Author is licensed under CC BY-NC-ND

The Background, current state, and

future of Artificial Intelligence

Artificial Intelligence is used widely today, it is used in everything

from data analysis to self-driving cars. In this report I shall follow
the development and effects of Artificial Intelligence from its
inception to present day in the various fields it is used in.

Ryan Gunn 26/11/23

Table of Contents
The Background, current state, and future of Artificial Intelligence...............................................1
Introduction.......................................................................................................................................3
Background of Artificial Intelligence...........................................................................................3
Can Machines Think?...................................................................................................................3
Dartmouth Summer Research Project on Artificial Intelligence.......................................4
Promising events and subsequent setbacks................................................................................5
Hardware and software Advancements....................................................................................5
Early AI and Machine Learning Advancements......................................................................5
Newell and Simon’s General Problem Solver......................................................................6
Joseph Weizenbaum’s ELIZA...................................................................................................6
Expert Systems...........................................................................................................................8
The Setbacks..................................................................................................................................8
The first AI Winter....................................................................................................................9
The second AI Winter..............................................................................................................10
AI Today.............................................................................................................................................10
Computational Power.................................................................................................................10
Availability of Data.......................................................................................................................11
Artificial Intelligence usage today............................................................................................11
The Problems of AI Today..........................................................................................................13
Ethics..........................................................................................................................................13
Model Autophagy Disorder...................................................................................................14
The Future of AI...........................................................................................................................14

Ryan Gunn 26/11/23

 Introduction
Artificial Intelligence (AI), Is a digital computer’s ability to think
for itself and carry out intellectual tasks regularly associated with
intelligent beings, such as the ability to learn from experience,
reason, generalize, problem-solve, have abstract thought and
discover meaning.

Although digital computing power and Artificial Intelligence

research and development has been increasing at a rapid pace
since the mid-1900’s, no Artificial Intelligence program has, just
yet reached the level of flexibility humans have over an array of
subject. AI, however, can transcend most human performance
within certain specific areas. (Shimek, 2023)

Background of Artificial Intelligence

Can Machines Think?

What is considered to be the earliest work in Artificial

Intelligence is the paper “Computing Machinery and Intelligence”
by Alan Mathison Turing, in which he proposes a question: “Can
machines think?” (Turing, 1950) In the first part of the paper,
Turing describes a game; the Imitation Game.

The Imitation Game Consists of three participants: an

interrogator, and two players, a man(A) and a woman(B). The
Interrogator is placed in a room separate from the two players’

Ryan Gunn 26/11/23

and all communication between two rooms are done through a
medium such as keyboards and a display screen.

The goal of the game is for the interrogator to correctly determine

which of the players is a man and which one is a woman. Player
A’s objective is to lead the interrogator to a false identification,
player B’s objective is to help the interrogator make a correct
identification. Turing then ponders the question: ‘What will
happen when a machine takes the part of A in this game?’.

Turing later states in the paper if a satisfactory amount of

interrogators were lead to a false identification by the computer it
is then to be considered as an intelligent, thinking being. The
imitation game was later renamed to “The Turing Test”.

Dartmouth Summer Research Project on

Artificial Intelligence

Early in the 1950’s there were several names for the field of
artificial intelligence, however, the name “artificial intelligence”
did not exist yet, and as such the field took various names such as
“thinking machines” and “automata theory”. It wasn’t until 1955
when John McCarthy decided to organize a group to clarify and
develop ideas on the subject. He picked the name ‘Artificial
Intelligence’ for the field.

After gathering funding for the project, the Dartmouth Summer

Research Project on Artificial Intelligence (DSRPAI) was ready to
proceed, this research project was a summer workshop which is

Ryan Gunn 26/11/23

widely considered to be the founding event of artificial
intelligence as a field. The event lasted for six to eight weeks.
During this event, several complex theories and ideas
surrounding artificial intelligence such as automatic computers,
self-improvement, neuron nets, abstractions and randomness
and creativity were fleshed out. (McCarthy, 1955)

Promising events and subsequent

setbacks
Hardware and software Advancements

From roughly 1956 to 1980 technological advancements surged,

some of the first commercial high level and object-oriented
programming languages emerged: Fortran and Simula (although
the first high level language was created in 1945 (Knuth & Pardo,
1976)). Computer computational power increased immensely
with the invention of the microchips, semiconductor memory –
which brought on the invention of RAM, SRAM and DRAM. The
CDC 6600 and Cray-1; the worlds’ first supercomputers were
developed and a whole host of other advances were made.

Early AI and Machine Learning

Advancements

Alongside the heaps of hardware and software advancements, AI

and machine learning flourished. Both theories and genuine
models were developed such as Newell and Simon’s General
Problem Solver (Aaron, 2023), Edward Feigenbaum’s Expert

Ryan Gunn 26/11/23

System: Drendal (Zwass, 2016) and Joseph Weizenbaum’s ELIZA
(Weizenbaum, 1966)

Newell and Simon’s General Problem Solver

Newell and Simon’s General Problem Solver (GPS) was a
computer program designed to emulate how a human being
would solve problems. It was written by Herbert Simon, John
Shaw and Allen Newell. GPS operates with 2 components; the
“core” of the program, which consists of how GPS would solve a
problem. The other component of the program is the learning
programs which would learn how to characterize the
environment that GPS is trying to solve. (A. Newell, 1959)

Joseph Weizenbaum’s ELIZA

Joseph Weizenbaum’s ELIZA was a model designed as a
psychotherapist, it worked off of a natural language processing
and allowed natural conversation with a computer. It was one of
the first natural language processing programs and thus one of
the first chatbots. (Weizenbaum, 1966) ELIZA was one of the first
programs ever able to attempt the Imitation Game (Turing Test).
(Cameron Jones, 2023) Below is a conversation with ELIZA;
“Men are all alike.
IN WHAT WAY
They're always bugging us about something or other.
CAN YOU THINK OF A SPECIFIC EXAMPLE
Well, my boyfriend made me come here.
YOUR BOYFRIEND MADE YOU COME HERE
He says I'm depressed much of the time.

Ryan Gunn 26/11/23

I AM SORRY TO HEAR YOU ARE DEPRESSED
It’s true, I am unhappy.
DO YOU THINK COMING HERE WILL HELP YOU NOT TO BE
UNHAPPY
I need some help, that much seems certain.
WHAT WOULD IT MEAN TO YOU IF YOU GOT SOME HELP
Perhaps I could learn to get along with my mother.
TELL ME MORE ABOUT YOUR FAMILY
My mother takes care of me.
WHO ELSE IN YOUR FAMILY TAKES CARE OF YOU
My father.
YOUR FATHER
You are alike my father in some ways.
WHAT RESEMBLANCE DO YOU SEE
You are not very aggressive but I think you don’t want me to
notice that.
WHAT MAKES YOU THINK I AM NOT VERY AGGRESSIVE
You don’t argue with me.
WHY DO YOU THINK I DON’T ARGUE WITH YOU
You are afraid of me.
DOES IT PLEASE YOU TO BELIEVE I AM AFRAID OF YOU
My father is afraid of me.
WHAT ELSE COMES TO MIND WHEN YOU THINK OF YOUR
FATHER
Bullies.

Ryan Gunn 26/11/23

DOES THAT HAVING ANYTHING TO DO WITH THE FACT
THAT YOUR BOYFRIEND MADE YOU COME HERE”
(Weizenbaum, 1966)
The capitalized lines are ELIZA responding. (Weizenbaum, 1966)

Expert Systems

Expert systems are AI models designed to emulate experts in

whichever domain the expert system is designed for. The expert
system gains its knowledge through real life experts inputting
their knowledge into its knowledge base, hence the name. The
system can then take an input situation, reason through its
knowledge base using its inference engine and return decisions
for that situation. The first expert system was developed by
Edward Feigenbaum in 1965, named the U.S. Dendral, Its purpose
was to analyse and identify chemical substances. Expert systems
have applications in numerous domains. (Zwass, 2016) (Orakzai,
2023)

The Setbacks

Following these events of advancements and improvements in

hardware and AI, the field continued to advance with newer and
newer projects, models and theories being developed year after
year, such as: Shakey the robot; the first ever “general-purpose”
robot (Abby, 2023),Mycin, an expert system for identifying the
cause of infections and recommend antibiotics for treating the
infection (Bales, 2023), Freddy robot; a robot designed to perform

Ryan Gunn 26/11/23

tasks at the same level as a five-year-old. The robot was created by
AI researchers at the University of Edinburgh and utilised two
camera eyes and a pressure sensitive grip. Although Freddy never
reached the level of intelligence of a five-year-old, it was still the
most sophisticated robot of its time (unknown, 2023).

The first AI Winter

Despite these advancements, it did not last. Around 1974 The first
“AI Winter” (AINewsletter, 2005) was beginning. The initial event
that caused the beginning of this winter is believed to be the
Lighthill report (Agar, 2020). The Lighthill report (officially titled
“Artificial Intelligence: A General Survey” (Lighthill, 1972)) was
commissioned by the British government to evaluate the state of
artificial intelligence research. The report had determined that
“AI had failed to achieve its lofty objectives and was unlikely to do
so in the foreseeable future.” (AINewsletter, 2023). Funding for AI
research in the UK by the government had been dramatically
reduced causing both projects and research to be left incomplete.

Even though this event only directly affected AI research in the

United Kingdom, Similar events happened in the United States of
America; the Defense Advanced Research Projects Agency
(DARPA) who was a major investor in AI research had cut the
funding for it due to growing scepticism.

This AI winter lasted around 6-7 years where both only little
research and projects were completed. (AINewsletter, 2023)
(Schuchmann, 2019)

Ryan Gunn 26/11/23

 The second AI Winter

Not even a full ten years had passed after the end of the first AI
winter for the second winter to be underway. After the end of the
first winter, the field of AI had shifted its attention to Edward
Feigenbaum’s expert systems, more precisely the practical
application of such systems. However, as people realised the
ineffectiveness of expert systems when pitted against a problem
they had never been trained on, once again the interest in AI
plummeted and as such the next winter took place.

AI Today
Today, AI is flourishing, both interest and funding into the field is
extremely high. The reason for AI flourishing today is the effect of
many factors but I believe two of the main driving factors are: The
computational power of modern computers and the availability
of data.

Computational Power
We have reached a point in computer hardware technology where
it is feasible to support large AI models. I believe this to be
because of Moore’s law, which is an observation that states that
the computational power and memory capacity of computers
doubles roughly every two years. Moore’s law has held up since its
conception in 1970.

Ryan Gunn 26/11/23

 Transistors per microprocessor
100000000000.00
10000000000.00
1000000000.00
# of Transistors

100000000.00
10000000.00
1000000.00
100000.00
10000.00
1000.00

1975
1977
1979

1985
1987
1989

1993
1995
1997
1999

2003
2005
2007
2009
2011

2019
1971
1973

1981
1983

1991

2001

2013
2015
2017

2021
Year

(Rupp, 2022).

Availability of Data

During the start of AI as a field - actually, before the widespread

use of the world wide web gathering large amounts of data would
have been extraordinarily hard. With as much as 30% of the
world’s population using the internet in 2010
(data.worldbank.org, 2021), there was already tonnes upon tonnes
of data for AI researchers to use to be able to create better and
more sophisticated AI and ML models.

As we are in an age of big data and more and more of the

population is using the internet each year (up to 63% in 2021
(data.worldbank.org, 2021)) I can only conclude that AI models
will just get better and better as long as researchers get smarter
about how the models are trained and programmed and do not
just rely on the sheer amount of data they have to train on.

Artificial Intelligence usage today

Ryan Gunn 26/11/23

With the world seeing how powerful AI can be if leveraged
properly, both companies and governments are investing into the
field. The UK has planned to become a global leader in AI
technology, in 2021 the UK had set forth a ten-year plan to do so,
alongside that, the UK has promised £500 million in November of
2023 for computing power for AI models, which brings the total
amount the country has invested in AI up to £1.5 billion. On top
of the investment by the government, Microsoft has pledged £2.5
billion into the UK in order to help the growth of the country into
an AI leader. (gov.uk, 2021) (gov.uk, 2023) (Landi, 2023)

Together with corporations, governmental bodies and academic

institutions across the world, AI and ML models are being used in
countless domains such as;
 Healthcare, to aid in diagnosis and treatment.
 The automotive industry, in order to help reduce accidents.
 The agricultural domain, to help increase productivity and
efficiency of farms.
 Software development, AI models can be used to increase
the speed of software development.
 The Cybersecurity sector, helping to ensure systems and
networks are secure.
 Archival; when writing this report, I had to reference a fair
few articles and reports from forty or fifty-plus years ago. I
was able to access these articles and reports because AI and
ML (more accurately computer vision and natural language
processing) models were used to archive them digitally.
 Translation; full
Beside these few that I have mentioned, AI is currently being
used in an absolute multitude of applications, way too many to
mention in this report.

Ryan Gunn 26/11/23

 The Problems of AI Today

Ethics

The first problem I am going to talk about is not exactly a

problem, it is a responsibility for AI programmers. When training
AI the programmers must be especially conscious about things
like bias, privacy and security. Programmers need to watch out for
bias because if an AI model was biased against or for certain
people or situations or whatever the model was intended to
analyse, it would most likely not be performing the job it was
designed to do and it is just not ethical to have a model be
prejudiced.

Privacy and security are another big ethical concern for AI

programmers as a lot of AI models use an obscene amount of
data, a large amount of which being personal data. Of course, if
any of this data is ever stolen by hackers, released on accident by
a programmer or the AI model spews out some of the personal
data when prompted, either on purpose or accidentally that
would be a massive breach of privacy.

There are also ethical concerns surrounding the overall use of AI,
such as, if deepfakes (deepfakes being the artificially created
images or video either by face swapping an already existing body
or a complete creation of the image or video to give the illusion
that someone was doing something that they had not. The term
“deepfake” is also used to describe the act of using AI to replicate
somebody’s voice.) using someone’s likeness are actually ethical

Ryan Gunn 26/11/23

uses of AI and not in breach of their privacy or putting
people(such as actors, voice-actors) out of jobs.

Model Autophagy Disorder

Model Autophagy Disorder also known as Model Collapse. This

“disorder” denotes a circumstance in which a ML model produces
repeated or low-quality results. This happens if the data it is
trained on was generated by an older model.
Any biases or mistakes/imperfections get amplified generation
after generation until the results keep repeating themselves.
Model collapse is especially a problem for models which rely on
massive amounts of data gathered through web scraping and
similar means, As well as models whose training data is not
sufficiently refreshed with fresh data. (Sanusi, 2023)
Models most affected by this are LLM’s like ChatGPT and Art
generation models such as Dall-e 2.

Although these are not the only problems plaguing the Field of
AI, I believe them to be the two most problematic.

The Future of AI

From all the research I have done when creating this report, I
have come to the conclusion that in the future, AI models will be
able to completely surpass human ability and I do not believe it
will be too far in the future either. If AI continues to advance at
the rate it is currently, I believe even 10 years in the future the
ability of AI will be completely unrecognizable from AI of today.

Ryan Gunn 26/11/23

I do not believe it is particularly hard to see that AI is going to be
incorporated in everything in the future, from transport to
healthcare to security - with a lot less oversight than it has today.
It is not exactly a bad thing; it may very well improve quality of
life for most of the population.

In the future I can predict that AI will take a much, much larger
role in cybersecurity as it will both be used to protect systems
from attacks and be used with the intention of breaking into said
systems.

Conclusion

Despite Artificial Intelligence’s rocky start with little or no

interest from the public and only a few scientists showing
interest, it survived and in the modern world, with both
governments and companies putting billions of pounds into AI
research, it is flourishing. I believe it is not unlikely that even a
few years in the future, artificial intelligence models will improve
immensely and surpass human ability.

Bibliography
A. Newell, J. C. S. H. A. S., 1959. A VARIETY OF INTELLIGENT LEARNING IN A GENERAL
PROBLEM SOLVER, Chicago: The RAND Corporation.

Aaron, D. M., 2023. Allen Newell. [Online]

Available at: https://www.britannica.com/biography/Allen-Newell
[Accessed 9 1 2024].

Abby, 2023. Shakey the Robot Explained: Everything You Need To Know. [Online]
Available at: https://history-computer.com/shakey-the-robot/
[Accessed 16 January 2024].

Ryan Gunn 26/11/23

Agar, J., 2020. What is science for? The L2020ighthill report on artificial intelligence
reinterpreted. [Online]
Available at: https://www.cambridge.org/core/journals/british-journal-for-the-history-of-
science/article/abs/what-is-science-for-the-lighthill-report-on-artificial-intelligence-
reinterpreted/61B13B32988D6A8C58CF8AADD4777789
[Accessed 16 January 2024].

AINewsletter, 2005. An AI Alphabet. [Online]

Available at: https://www.ainewsletter.com/newsletters/aix_0501/#w
[Accessed 16 January 2024].

AINewsletter, 2023. The AI Winter Phenomenon. [Online]

Available at: https://www.ainewsletter.com/newsletters/ai-winter-phenomenon/
[Accessed 16 January 2024].

Bales, R., 2023. Mycin Expert System: The First AI Medical Diagnosis. [Online]
Available at: https://history-computer.com/mycin-expert-system-the-first-ai-medical-
diagnosis/
[Accessed 16 January 2024].

Bradshaw, H., 2022. Microsoft is hoping its algorithms can help farmers—and the planet.
[Online]
Available at: https://www.popsci.com/technology/microsoft-ai-agriculture/
[Accessed 21 January 2024].

Cameron Jones, B. B., 2023. Does GPT-4 Pass the Turing Test, Ithica: Cornell University.

data.worldbank.org, 2021. Individuals using the Internet % of population. [Online]

Available at: https://data.worldbank.org/indicator/it.net.user.zs?end=2022&start=1999
[Accessed 21 January 2024].

gov.uk, 2021. New ten-year plan to make the UK a global AI superpower. [Online]
Available at: https://www.gov.uk/government/news/new-ten-year-plan-to-make-britain-a-
global-ai-superpower
[Accessed 21 January 2024].

gov.uk, 2023. Boost for UK AI as Microsoft unveils £2.5 billion investment. [Online]
Available at: https://www.gov.uk/government/news/boost-for-uk-ai-as-microsoft-unveils-25-
billion-investment
[Accessed 21 January 2024].

Knuth, D. E. & Pardo, L. T., 1976. The Early Devlopment Of Programming Lnaugages, Stanford:
Stanford University.

Landi, M., 2023. Chancellor pledges £500m investment to boost UK AI development. [Online]
Available at: https://www.independent.co.uk/business/chancellor-pledges-ps500m-
investment-to-boost-uk-ai-development-b2451781.html
[Accessed 21 January 2024].

Lighthill, J., 1972. Artificial Intelligece: A General Survey, Cambridge: Chilton.

McCarthy, J., 1955. A PROPOSAL FOR THE DARTMOUTH SUMMER RESEARCH PROJECT ON
ARTIFICIAL INTELLIGENCE. [Online]

Ryan Gunn 26/11/23

Available at: https://web.archive.org/web/20070826230310/http://www-formal.stanford.edu/
jmc/history/dartmouth/dartmouth.html

Orakzai, R. U., 2023. What Are Expert Systems?. [Online]

Available at: https://www.baeldung.com/cs/expert-systems
[Accessed 15 January 2024].

Rupp, K., 2022. Github. [Online]

Available at: https://github.com/karlrupp/microprocessor-trend-data
[Accessed 15 January 2024].

Sanusi, T., 2023. What Happens When AI Eats Itself. [Online]

Available at: https://deepgram.com/learn/when-ai-eats-itself
[Accessed 21 january 2024].

Schuchmann, S., 2019. History of the first AI Winter. [Online]

Available at: https://towardsdatascience.com/history-of-the-first-ai-winter-6f8c2186f80b
[Accessed 16 January 2024].

Shimek, C., 2023. Neuro Science News. [Online]

Available at: https://neurosciencenews.com/ai-creativity-23585/
[Accessed 4 12 2023].

Turing, A. M., 1950. Computing Machinery and Intelligence. Mind, pp. 433-460.

unknown, 2023. Freddy the Robot. [Online]

Available at: https://www.nms.ac.uk/explore-our-collections/stories/science-and-technology/
freddy-the-robot/
[Accessed 16 January 2024].

Weizenbaum, J., 1966. Computational Linguistics. Communications of the ACM, 9(1), pp. 36-45.

Weizenbaum, J., 1966. Computational Linguistics. Communications of the ACM, 9(1), pp. 36-45.

Zwass, V., 2016. Expert Systems. [Online]

Available at: https://www.britannica.com/technology/expert-system
[Accessed 9 January 2024].

Ryan Gunn 26/11/23

Ryan Gunn 26/11/23