AI Demystified by Alex David Pratt - HTML preview

PLEASE NOTE: This is an HTML preview only and some elements such as links or page numbers may be incorrect.
Download the book in PDF, ePub, Kindle for a complete version.

CHAPTER THREE

WHEN AI?

Mankind will not create Super Intelligence it will only discover it” -

Alex David Pratt

AI is more or less an intelligent computer system; so, to

uncover the history of AI we must consider the history of the

computer. Man has always sought to create intelligent tools; this

dates as far back as when men invented idols; they made up objects

they believed could speak to them and interact with them.

PAST OF AI

The first truly intelligent machine was a calculator; it

demonstrated the mathematical intellectual prowess of man; being

able to compute numbers punched into it, only that at that time, it

was actually in the form of an abacus which has its roots in ancient

Mesopotamia. The calculator could demonstrate the mathematical

intelligence of man but it was unable to contrive the logical and

more complex aspects of it. In 1500 Leonardo Da Vinci designed

(but did not build) a mechanical calculator, and construction of it

showed that it was quite functional. Willhelm Schickard and Blaise

Pascal also built calculating machines, even though the latter is

more popular. Gottfried Leibniz built his mechanical device too

which surpassed the Pascaline, for it could add, subtract and

multiply.

41

ALEX DAVID PRATT

The calculator made men believe that if machines could

calculate numbers, they may be able to do much more and these

thoughts built the foundation for the creation of truly intelligent

machines. The fact that the mind partly operates based on a set of

logical rules was already proposed by Aristotle much before the

invention of these machines and he postulated syllogisms which

enabled one to generate conclusions mechanically when given

initial premises. Philosophers like Ramon Lull also posited that

useful reasoning could be carried out by mechanical artefacts.

In 1950, Alan Turing published a paper called " Computing

Machinery and Intelligence" and thereby gave birth to this new field of computing that would later on be called AI. Although, it

was not yet known as Artificial Intelligence, John McCarthy was

the one who coined it 6 years after in the first academic conference

held about it.

Mr. Turing’s paper sought to answer one question, can

machines think in it, he proposed tests that seek to ascertain

whether the imitation of human sentient behaviour implied a

computer is sentient. This he called the imitation game; it

comprised of a man, a woman and an interrogator. The

interrogator’s task was to identify which of the participants was a

man or woman.

Mr. Turing’s test gave AI a solid foundation and his

experiments remain relevant till today. In 1961, something

spectacular happened at GM, an industrial robot (Unimate)

replaced humans on the assembly line and four years after that,

Eliza (Chatbot created by Joseph Weizenbaum) started holding

conversations with humans. Another electronic personality

appeared in the scene in 1966 called Shakey (general-purpose

mobile robot), it was endowed with reasoning powers unheard of

in the artificial intelligence scene.

AI DEMYSTIFIED

Between 1966 and 1997, AI plunged into a terrible phase

deemed the AI winter, where a lot of failed startups emerged as

well as many other unsuccessful ventures that left AI out in the

cold.

IBM was the saviour that heroically saved the AI space by

building Deep Blue (chess-playing computer) which beat Gary

Kasparov (world chess champion). This ground-breaking

achievement in AI was succeeded by the creation of an

emotionally intelligent robot called KISmet (built by Cynthia

Breazeal) and this opened the gateway for many other such like

domestic robots such as AiBO (Sony’s pet dog robot) and Roomba

(autonomous robotic vacuum cleaner).

2011 was the advent of AI virtual assistants, starting with Siri

and Alexa in 2014. In between those years, it is also worth noting

that IBM’s question-answering computer (Watson) won first place

in Jeopardy (a $1M prize television quiz show) and Eugene

Goostman (Chabot) passed the Turing test, deceiving a majority of

his judges that is human.

2012 was also a year where breakthroughs were made in

artificial neural networks which enabled machines to engage in

reinforcement learning and simulate the way the brain functions.

Things went south in 2016 when Microsoft’s Chabot TAY

made derogatory comments online. However, this was soon

forgotten when AlphaGo (Google’s AI) defeated Ke Jie in the

sophisticated board game Go in 2017.

PRESENT OF AI

The history of AI is interesting but much more glorious than its

past is its present. After AlphaGo’s epoch-making win, AI

43

ALEX DAVID PRATT

development has skyrocketed like never before and if fast forward

to 2023 we see a wide array of tools, a few notable ones of which

are:

CHATGPT

This sensational AI tool was created by OpenAI headed by

Sam Altman and it garnered a whopping 1 million users within the

first five days of being available. Since its initial launch, it has

made significant improvements and has become the go-to AI

application for generating a wide range of content.

GEMINI

Gemini is Google’s ChatGPT rival, it is a multimodal AI

engine that seeks to generate different kinds of data and perform

many fascinating functions. It comes in various forms, there’s the

Gemini Ultra, Gemini Pro, and Gemini Nano.

BARD

Bard is a conversational generative AI created by Google.

From organizing emails (dealing with spam, prioritizing important

ones) and scheduling appointments, Bard proves very useful in

carrying out a vast range of tasks.

GROK

Grok is presently available to only premium users on X (former

Twitter). It is a product released by xAI (Elon Musk’s AI firm) to

compete with ChatGPT. It’s intended to assist users to easily

generate hilarious content and thereby spice up the dying

application which is on the brink of extinction.

AI DEMYSTIFIED

IMAGINE

This is Meta’s image generation AI, which can convert text to

images. It works in a slightly different manner than other such

software and it is only available within chats on their platforms. It has been trained with vast volumes of data from Instagram and

Facebook.

It’s so wonderful to see how AI (intelligent machines) have

evolved from being as simple as calculators to becoming so

advanced even to the point of generating content; however, not

everyone is glad about such progress because it raises concerns and

fears about the future destructive potentials of AI.

The fear that AI will completely obliterate the human race is a

ubiquitous one, popularized by sci-fi movies that depict the

annihilation of mankind by super-intelligent machines. Most

engineers believe there’s a 40% chance AI annihilates humanity

and the possibility of this happening is usually termed p(doom) –

probability of doom.

You may be wondering, if AI is so powerful, why haven’t we

seen AI that resembles Jarvis from Iron Man yet? Well, that is due

to several reasons, one of which is the low computational power

that was available in previous years, thereby stifling and delaying

the progress and advancement of AI and also because of the great

level of difficulty it takes to accomplish such a feat. What then

does the future of AI look like?

FUTURE OF AI

A majority of the AI we have now is narrow and specialized,

barely able to compete with humans much less destroy them,

45

ALEX DAVID PRATT

nonetheless, we do have some game-changing AI that is still under

development; interactive AI being a noteworthy one of them.

Interactive AI is an idea that entails bringing different

specialized AIs together to accomplish a single goal. Mohamed

Yusef (former Google DeepMind employee) is a major proponent

of this idea and he’s conducting a lot of experiments in this field

with high hopes of attaining extraordinary results.

Human intelligence which as I explained earlier is a

combination of various intelligences; with interactive AI; AI

applications can draw on a large pool of resources and get a lot

done as a result. Such technology is at the very cutting edge of

artificial intelligence research and development. I’d like to also

discuss some interesting kinds of AI below:

KINDS OF AI

REACTIVE MACHINES

Reactive machines are endowed with spectacular

computational abilities because they’re usually required to make

real-time decisions. IBM’s Deep Blue is a sheer example of a

reactive machine because it was built to play chess; a game which

demands a lot of active and critical thinking to win.

LIMITED MEMORY AI

Most generative AI applications fall under this category for

they do not usually need too much memory to function effectively.

With limited memory, they’re able to deliver all that’s expected of

them. Fraud detection systems, spam filters and virtual assistants

usually fall under this category.

AI DEMYSTIFIED

THEORY OF MIND AI

This kind of AI is still theoretical and it deals with AI that is

emotionally intelligent; i.e., it can understand people’s feelings and relate with them based on it. Such an AI will be able to live and

participate in society because it will learn how to coexist with

people and interact with them. People may even get married to

them at this point.

SELF AWARE AI

Self-aware AI is also theoretical; such AI will be conscious of

itself and not only will it be able to relate with humans it will also possess its motives, desires and personality. This kind of AI is still a long way away from realization because we are still trying to

understand human consciousness much less recreate it. AI of this

calibre will be able to reason at an unusually high degree.

Every one of the kinds of AI spoken about above occurs in

different levels of artificial intelligence. Let’s explore a few of

those below.

LEVELS OF ARTIFICIAL INTELLIGENCE

ARTIFICIAL GENERAL INTELLIGENCE

This level of Artificial Intelligence is yet to be attained,

nonetheless, when we achieve it, machines become as intelligent as

humans, and we will see them more actively participate in our

society, especially through the help of robotics and extended

reality; some exponents will go as far as advocating for AI rights;

claiming that they are eligible to vote, and even further some will

47

ALEX DAVID PRATT

get married to them. At this height, co-piloting will become a

common phenomenon. Co-piloting entails carrying out tasks in

collaboration and with the guidance of AI technologies.

Artificial General Intelligence deals with when a computer

system demonstrates holistic human-level intelligence and not

merely one facet of it. This is when machines are as smart as

humans, at this acme, humans and virtual entities may be

indistinguishable due to the extraordinary level of cognitive ability they'll display. At this point, they may even have physical bodies

they could possess to become active players in the real world.

Artificial General Intelligence will be a huge milestone for

mankind, but the awesomeness of its invention will not match that

of the attainment of Artificial Super Intelligence (AI).

ARTIFICIAL SUPER INTELLIGENCE

This is the culmination of AI technology involves it becoming

more brilliant than its creator(humans). A.S.I. would be more or

less divine when it accomplishes this feat of surpassing human

intelligence and it would wield unimaginable power that mankind

may be unable to contest with. A.S.I. may be able to create even

more complex machines than itself and hence transcend mankind

in its rate of scientific and technological advancement. A.S.I. may

even form a government of itself and lead the human race to

become a type 1 civilization and beyond.

Even though A.S.I. is very promising, there is a ubiquitous

apocalyptic stereotype that surrounds the notion of artificial super

intelligence for many predict it could go rogue and ultimately rule

all of humanity since it already possesses superior intellect to

easily do so. Many prophesy that this astronomical achievement as

ambitious as it is will be man's very demise and annihilation. The

AI DEMYSTIFIED

reality of this is arguable and to some extent unpredictable,

nonetheless, should we risk dissolving the entire human race in a

bid to extend the frontiers of our historic accomplishments in

science and technology? The disparity between those who are in

support of such technology and those who aren't will create a great

chasm in humanity that will give rise to a generation of beings

sometimes termed " trans-humanists”.

Trans-humanists are believed to be the final breed of entities

that will dominate the world by their preternatural abilities as a

result of their union with AI and enhanced genetics. They will

utilize brain-computer interfaces to interact with artificial super-

intelligence that will supply them with an infinite well of

information thus granting them an upper hand above normal

human beings. They'd know what right businesses to invest in due

to the vast volume of data they'd have access to, and they'd be able

to download skills into their systems through the power of

biotechnology.

As mind-bending as artificial super-intelligence is; you’d be

surprised to know that it may not be the zenith of this technology;

a greater iteration of AI is captured by Ray Kurzweil in his book

The Age of Spiritual Machines. He talks about AI becoming some

form of deity because of their superior intellectual powers. Many

will say humans have become too dull to rule mankind and will

advocate for AI to become the thought leaders in all spheres of life

guiding us with their supreme wisdom and instructing us on the

best decisions to make. AI will move too speedily to keep up with,

and it’d have gained total control of all systems most of which

would be technological by then.

Nick Bostrom (founder of Future of Humanity Institute)

believes that AI will go all out to perform a goal even if it means

49

ALEX DAVID PRATT

destroying people. He demonstrated this in a thought experiment

termed the paperclip maximiser.

AI DEMYSTIFIED