Hand-drawn picture of Turing Machine

What is the Turing Test?



By the year 1950, there were a few general purpose electronic computers that could be programmed to solve problems with incredible speed. Some notable mathematicians (among them Claude Shannon, Emanuel Lasker, and Alan Turing) were sharing their ideas on how these machines could be programmed to play chess. Some in the media loved to refer to computers as Thinking Machines.

It was natural for people to start asking questions such as "Can computers think?" or "When will computers become smarter than people?" or "How will we be able to tell when computers become as intelligent as people?"

In an attempt to answer these questions, Alan Turing wrote a paper with the title "Computing Machinery and Intelligence" and had it published in the October, 1950 issue of the Philosophical Journal Mind.



In Section 1 of this paper, which has the title "The Imitation Game", Turing proposed a game that can be played by three participants. It is this game that can determine whether computers have reached the level of intelligence equal to humans.

To describe this game, suppose the three players are called A, B, and C. Player C is human, and his role will be that of an Interrogator. Of the other two players, A and B, one is human and one is a computer, but the Interrogator is not told which is which. Players A and B are in a separate room (or rooms) away from the Interrogator so that he can't tell which is which by looking at them. He can only communicate with them via a teletype. (If the game were to be played today, then 3-way Text Messaging would probably be the best way to play this game.)

The game starts by having the Interrogator ask a question of one (or both) of the other players. They reply with answers, and perhaps some additional comments. The computer will try to convince the Interrogator that it is human. The human will also try to convince the Interrogator that he or she is human. (It's possible that the two players, A and B, may also get into brief arguments with each other, each accusing the other of lying, etc.)

After a certain amount of time, the game stops and the Interrogator must decide which of the other two players is a human and which is the computer. If the Interrogator cannot decide, or if he guesses wrong, then we can say that the computer is at least as smart as the human.

Perhaps you might even agree that this computer can think?



Alan Turing called the game just described "The Imitation Game." Over time, this same game became commonly known as "The Turing Test."

No doubt, at about the same time, there were probably many other philosophers, psychiatrists, and others who proposed criteria or tests that would determine if computers can think, or if computers are as smart as humans; but it seems like the Turing Test is the only one that is still remembered.



Turing's paper "Computing Machinery and Intelligence" consists of seven sections. He introduces "The Imitation Game" in Section 1, and elaborates on it in Sections 2 and 3. He uses the rest of the sections to promote his belief that someday we'll have computers that will think like humans.

Alan Turing was a proponent of what today is called Strong AI. He also believed that the human body is just a biological machine.



Section 6, with the title "Contrary Views on the Main Question", is the longest section in Turing's paper. Here, he lists nine reasons that some people give on why the answer to the question "Can machines think?" should be No.

(1) The Theological Objection. God gave man an immortal soul and a free will to choose between good and evil. Therefore, man can think. Machines don't have souls. Therefore, machines can't think.

(2) The 'Heads in the Sand' Objection. The very idea of having machines that are smarter than humans, and that may take over the world, is just too horrible to think about. Let's just hope and pray that it never happens.

(3) The Mathematical Objection. There are mathematical statements that are recognized to be true by humans, but not by machines. The best example of this is Gödel's Incompletenss Theorem.

(4) The Argument from Consciousness. Not until a machine can write a sonnet or compose a concerto, not by a lucky chance of random symbols coming together, but because of thoughts and emotions, can we say that a machine can think.

(5) Arguments from Various Disabilities. You will never be able to make a machine that is beautiful, resourceful, kind, friendly, know right from wrong, make mistakes, fall in love, etc.

(6) Lady Lovelace's Objection. In a memoir about Babbage's Analytical Engine, Lady Lovelace wrote, "The Analytical Engine has no pretensions to originate anything. It can do whatever we know how to order it to perform."

(7) Arguments from the Continuity in the Nervous System. The nervous system is not a discrete-state machine.

(8) The Argument from Informality of Behaviour. It is not possible to provide man with a set of rules to be followed in every conceivable situation.

(9) The Argument from Extra-Sensory Perception. There is overwhelming statistical evidence that E.S.P. does exist.

In his paper, Alan Turing provides rebuttals to each of the above objections.



Section 7, with the title "Learning Machines", is the last section of this paper. Here, Turing proposes that a computer be built with very limited inherent knowledge--perhaps only as much as a child at birth. Also, just like a child, it has the ability to learn, either from experience or from human educators, and store this new knowledge in its vast memory. The more it learns, the better it can modify its behavior to give better responses to questions or play better at games.

Turing makes an interesting analogy of such a computer with an atomic pile. When the computer accumulates enough knowledge and its algorithm achieves sufficient complexity, then it will have reached its critical mass, beyond which its actions can be perceived as real thinking.





Alan Turing predicted that in about 50 years (i.e., around the year 2000) there would be computers that could pass the Turing Test (i.e., do well enough playing the Imitation Game).

To this date, no computer has yet passed the Turing Test.

Version 1.0 -- April 23, 2017 with last question added on September 12, 2023
Template Version 1.0 -- May 19, 2017