Computing Machinery and Intelligence, written by Alan Turing more than half a century ago (in 1950), is a seminal paper about the topic of Artificial Intelligence. The paper introduced a concept of what is now known as the Turing Test for determining if a machine is considered intelligent.
Computing Machinery and Intelligence concerned about the question “Can machines think?” Because the meaning of the words “machine” and “think” may be unambiguous, Turing proposed to replace the question with a more specific question: “Can machines win the imitation game?” Definition of the imitation game and supporting reasons is given later in the paper.
The Imitation Game
By playing the imitation game, as the name suggests, a machine has to imitate human’s behavior so that it cannot be distinguished from a human.
There are three participants involved in the game: a man (A), a woman (B) and an interrogator (C) who may be of either sex. They are in isolated rooms and not allowed to see, touch or talk to one another. Communication is done via a telepath printer (that is by typewritten texts). The interrogator asks questions and the man and woman answer. The interrogator’s object is to determine who the man is and who the woman is. The man’s object is to pretend to be the woman to trick the interrogator. The woman’s object is helping the interrogator by trying to show the truth that she is a woman.
After describing the game, Turing asked, “What will happen if a machine takes part of A in this game?” That is, the machine should try to pretend to be a human so that the interrogator decide wrongly as often as A is played by a human.
Turing argued that if the machine plays so well that the interrogator decides wrongly, it should be considered intelligent.
Critique of the new problem
Turing described a strong objection to the Imitation Game: it is weighted too heavily against the machine (conversely, if the man were to pretend to be the machine he would clearly make a very poor showing). Could machines carry out something, which should be described as thinking, but which is very different from what a man does?
Turing also concerned about which kinds of machines we should consider. He points out that machines should be digital computers instead of things like clones (while still man-made). (Digital computers are machines that manipulate binary digits and write it into the memory using simple rules). Turing supported the use of digital computers because they already existed at the time the paper is written. Another reason is that digital computers are “universal”.That is, if a digital computer can do something, every sufficiently powerful digital machine can.
Contrary views to the main question
After clearly stating the question and its conditions, Turing discussed about its nine common objections. He noted that the original question “Can machines think?” would be too meaningless to deserve a discussion but cannot be neglected.
1. Theological Objection
The objection states “Thinking is a function of man’s immortal soul that God has given. Hence no machine can think”.
“In attempting to construct such machines,” Turing opposed this objection in theological terms, “we should not be irreverently usurping His power of creating souls, any more than we are in the procreation of children: rather we are, in either case, instruments of His will providing mansions for the souls that he creates.”. At the end, he said, “With our present knowledge, such an argument appears futile”.
2. ‘Heads in the Sand’ Objection
The objection states, “The consequences of machines thinking would be too dreadful. Let us hope and believe that they cannot do so.”
Turing gave his points that this objection was connected to the theological argument and was not substantial enough to require refutation.
3. Mathematical Objections
Based on mathematical results such as Gödel’s incompleteness theorem and results in theory of computation by Turing himself, the objection states that there are limits that computers can never achieve.
Turing argued that although there are limitations to the power of computers, no such limitations apply to the human intellect. Moreover, he believed that those with Mathematical objections would mostly accept the imitation game as a basis for discussion.
4. Argument from Consciousness
This objection is based on the belief that machines can never experience emotions or feelings like a human can.
Turing replies by saying that we have no way to know if any individual other than ourselves experiences emotions. He then provided a convincing example that helps persuade people with this objection.
5. Arguments from Various Disabilities
This objection states there are many things that a machine will never be able to do. For example, be kind, resourceful, beautiful, friendly, have initiative, etc.
Turing noted, “No support is usually offered for these statements” and chose a few things to answer:
Machines cannot make mistakes: We can program a machine to pretend to make mistakes.
Machine cannot be the subject of its own thought: Turing said, “A machine undoubtedly can be its own subject matter. It may be used to help in making up its own programmes, or to predict the effect of alterations in its own structure. By observing the results of its own behaviour it can modify its own programmes so as to achieve some purpose more effectively”.
Machines cannot have much diversity of behavior: Turing notedthat, with enough storage capacity, a computer could behave in avery large number of different ways.
6. Lady Lovelace’s Objection: Machine cannot originate anything. It can only do what we order it to perform.
Turing replied that computers could still surprise humans when the consequences of different facts are hard to recognize.
7. Argument from continuity in the nervous system:
The nervous system is continuous. Therefore, a discrete-state machine like a digital computer can never mimic the behavior of the nervous system.
Turing argued that discrete-state systems could simulate analog systems to a reasonable degree of accuracy if it is given enough computing power.
8. Argument from the informality of behaviour:
We can predict the future behavior of a discrete-state machine because it is governed by laws. Therefore, it is not truly intelligent.
Turing replies to this objection by stating that the machine behavior would become very difficult to predict.
9. Argument from Extra-sensory perception (ESP):
Human have ESP such as telepathy, clairvoyance, etc that are still beyond scientific light. Therefore, machines have difficulty, if not impossible, to mimic ESP behavior of human.
Turing suggests that we can control ESP by something such as the telepathy-proof room.
Turing proposed to create a learning machine that imitates a child’s mind rather than an adult mind. He believed that there is little mechanism in the child brain so that it could be easily programmed.
The machine consists of two parts, the child program and the education process. The learning process may be regarded as a search for a form of behaviour, which will satisfy the teacher.
“We can only see a short distance ahead, but we can see plenty there that needs to be done”, said Turing. 50 years has passed, Turing’s paper is really an important point in the field of Artificial Intelligence.