⌈Computing Machinery and Intelligence⌋, a summary for those who don't understand like me

...written by a 15 year old

Not so brief summary and important notes

The subtitles of this part matches the subtitles of Turing's paper.

(1) The imitation game

Alan Turing starts by introducing a new game, the imitation game. This game is meant to help answer the question, 'Can machines think?'

He then explains another form of the question, but less ambiguous, to help the reader understand the imitation game.

(A) is a man, (B) is a woman, and (C) is the interrogator.

• (C) is to try guess which person either (A) or (B) is of either sex man or woman
• (B) is to help the interrogator guess correctly

Another important thing here is that the interrogator should be communicating only using a typewriter. Using a typewriting can hide the person's voice, pitch, handwriting and more.

But the interrogator won't know whether (B) is telling the truth or not.

Now, replace (A) with a machine and the interrogator will try to find out whose the machine. In short, Alan Turing's opinion is that if the interrogator is wrong on 70% of the time, the question is meaningless to answer because it already suggests machines are smart enough to fool another human into thinking that it is a human.

(2) Critique of the New Problem

The game is to hard on the machine. The interrogator could simply ask to perform a hard arithmetic problem and the individual to answer correctly will be the machine. So, can't machines do things the best to their ability rather than act like a human? Turing ignored this problem, saying that it could go away if we set the machine to play the imitation game satisfactorily. This suggests that Turing emphasizes the importance of AI acting like a human rather than acting rationally.

In today's world, acting rationally is emphasized and generally considered to be the definition of AI (though arguable).

Another interesting text to point out is Turing thinks ating like a human is not the best strategy for the machine. Not sure why Turing thinks so and how the machine would win if they weren't acting like a human.

(3) The Machines concerned in the Game

Turing gives 3 conditions for machines to ideally have.

1. We should allow every kind of engineering technique to be used
2. We should allow the possibility of experimental machine (so experimental, in fact, that the engineers don't even know what is going on), which is exactly what AIs are like today
3. The machines shouldn't be born naturally, but made artificially

(4) Digital Computers

Turing introduces one of the most oldest planned machine, the Analytical Engine. This engine is planned out by Charles Babbage, a mathematician in Cambridge. Turing notes that unlike in modern machines, Babbage planned on using purely mechanical parts as storage. He planned on using wheels and cards for storage instead of electricity.

Turing writes the following:

The fact that Babbage's Analytical Engine was to be entirely machanical will help us to rid ourselves of a supersition. Importance is often attached to the fact that modern digital computers are electrical, and that the nervous system also is electrical. Since Babbage's machine was not electrical, and since all digital computers are in a sense equivalent, we see that this use of electricity cannot be of theoretical importance ...(shortened)... The feature of using electricity is thus seen to be only a very superficial similarity. If we wish to find such similarities we should look rather for mathematical analogies of function.

Summarizing, Turing thinks that machines using electricity is only a superficial similarity and that it is not of theoretical importance. Not sure what he means, but if he means that mechanical machines can be smart, he seems to be wrong on this one as 100% of AI today runs on and saves data on electricity.

(5) Universality of Digital Computers

He then introduces 'discrete state machines'. Basically, discrete state machines are machines that are programmed to work for specific sets of scenario.

He points out an important property of this machine. This kind of machine is predictable and has the ability to mimic other machines. So, what if the smart machine plays one role and the mimicing machine plays the other role? The interrogator won't be able to notice the difference between these two.

He points out these kind of machines, if programmed to mimic the smart machine, can play the imitation game, too. He did point out that this machine would need adequate amount of storage and fast speed. But Turing, again, seems to be wrong on this one. There is almost no way to program every scenario as it would take more storage than ever imagined and the machine wouldn't be able to take action in unprogrammed scenario.

Here, the 'universality of digital computers' means the property to copy other discrete state machines.

(6) Contrary Views on the Main Question

Turing's Prediction

He gives a prediction in 50 years (2000):

1. He predicts that there would be machines programmable in a storage of 1GB. Back then, 10KB was widely used (history of data storage)
2. He predicts that with 1GB, the machine would be able to play the imitation game, tricking the interrogator more than 70%.

He says if this is achieved, 'Can machines think?' question would be meaningless to ask.

But, boy was he wrong on this prediction.

1. IBM achieved 50GB.
2. The first chatbot to pass the Turing Test, without any workarounds, was Google’s LaMDA around 2022.

Opposition Views

(1) The Theological Objection

• Summary of Objection: God has given soul to only man. God did not give soul to animals or machines. Hence, animals or machines cannot think.

(2) The 'Heads in the Sand' Objection

(3) THe Matematical Objection
(4) The Argument from Consciousness
(5) Arguments from Various Disabilities
(6) Lady Lovelace's Objection
(7) Argument from Continuity in the Nervous System
(8) The Argument from Informality of Behaviour
(9) The Argument from Extra-Sensory Perception