Summarising The Imitation Game by Alan Turing

1. The Imitation Game

• A subtle introduction to the very concept of machine based intelligence by posing various arguments that question the concept itself.
• Instead, we try to frame the same concept in a different way by laying foundation for a game which consists of three people namely a man (A) a women (B) and an interrogator(C) who may be of the either sex. C has to figure out which category A, B belongs to given two categories namely X,Y.
• A’s objective is to confuse C for every question asked by C while B is there to help C. Now if we replace A with a machine, what kind of changes in approach of C can be expected and how quickly will he be able to find out?

2. Critique of the new problem

• It is important to understand and consider the distinguishable line between a man and a machine and how we can reduce the gap or what is our criteria to decide the line. While some examples reveal the differences, others can help close the gap.

3. The Machines considered in the game

• We are more interested in a machine which can replace A, and in the current ecosystem, digital computers are the one which can assume the required role.

4. Digital Computers

• A computer which can store information, has an execution unit to perform individual operations involved in a calculation and a control unit which orchestrates the execution steps.
• Main objective behind digital computers has always been to mimic human actions quite closely through a process that we call programming.

5. Universality of the digital computers

• Universality in this context mostly refers to the ability of digital computers to take up the role of any discrete state machine present currently in the world because of their storage capacity and processing capability. This can push digital computers a bit further to assume the role of a person in the game. (Note this is him writing in 1950s about computers)

6. Contrary views on the main questions

• Machines are not similar to humans on intellectual grounds since humans have soul/consciousness. And hence a conflict of thought occurs when we are interested in understanding how god has control of beings and his ability to confer them with soul or intellect while he cant have any control of things made by man(such as numbers).
• We as humans tend to feel subconsciously that we are the most superior form of race and no one will be able to replace our intellectual capabilities.
• Expression of thoughts, ideas, emotions etc is something that widens the gap further. But arguably that’s a solipsistic point of view since we cannot understand how a particular human is expressing or feeling. Suppose we have a machine in place and it was able to answer questions from a sonnet. Consciously we tend to refrain from the test considering it worthless against a machine at least on the former criterion.
• Usually as a human, if at all a machine is to take the place of A and has to help interrogators categorize him as X, he needs to prove that he possesses the features of X and supposedly he should be having typical human characteristics.
• Another form of argument is that “Machines can never be innovative”, which is arguably due to lack of understanding of terms innovation and creativity which is often reciprocated in terms of cumulation of all possible consequences corresponding to a fact.
• As a matter of fact, a discrete system cannot comprehensively be compared to a nervous system which has a continuous nature. But a fair comparison between a differential analyzer(a continuous system) and a digital computer can help us narrow down the gap.
• If the interrogator by means of telepathy is able to control the machine’s ability to give an answer, there would always be an associated probability in the machine’s response if the machine is programmed to give the correct but random response(as an example).

7. Learning Machines

• A machine’s capability of delivering new ideas and perspective may seem vague as per Lady Lovelace’s objection. But if we pick up ideas from a human child as in how it is trained intellectually and becomes a thinker once in the adult phase, we shall be able to do the same with a child machine. (Lady Lovelace’s Objection: One of the most famous objections states that computers are incapable of originality.)
• We can have either a simple or a complex inheritance for the child machine consisting of well-established facts, conjectures, mathematically proved theorems, statements given by an authority, expressions having the logical form of proposition but not belief-value. In addition, substantial ignorance by the experimenter and nature of randomness can help the child machine to evolve and learn better.
• Maybe we can start with a game of chess or teaching a particular language.

Thank you for reading :)