Skip to content
Coordinates of ThoughtHome of Philosophy
Join

Turing test

Updated 19 November 2026

On this page
  1. The argument at a glance
  2. What it means
  3. The questions it belongs to
  4. Thinkers
  5. To read
  6. Sources and further reading
  7. About this page
  8. Cite this page

Alan Turing's 1950 proposal to replace the question 'Can machines think?' with a game in which an interrogator, conversing by text, tries to tell a machine from a human; passing is offered as evidence of intelligence.

Also: imitation game

The argument at a glance

A guide to the entry below, quoting its own words. Skip to the full entry and its sources.

Question
Can the question whether machines can think be settled at all, and if so, what kind of evidence would settle it? Turing's essay asks whether the ordinary question can be replaced by one that admits of precise discussion, and whether the objections to machine thinking defeat the replacement.
Reasons , COT reconstruction: the numbering is ours, not the philosopher's

(COT's reconstruction; Turing does not number his steps)

  1. The question 'Can machines think?', as ordinarily posed, is too unclear to be discussed profitably.
  2. The question whether a digital computer could do well in the imitation game is precise and can be investigated.
  3. If a machine did well in the game, we would have as much reason to think that it thinks as we have to think that other people think; this premise is drawn from Turing's reply to the argument from consciousness.
  4. None of the objections Turing considers shows either that a digital computer could not do well in the game or that doing well would not count.
  5. So the precise question may stand in for the original one, and the original need not be answered first.
Intended conclusion
The question whether machines can think should be approached through the question whether a machine could do well in the imitation game, and Turing expected the answer to the second question to be yes. On the Stanford entry's reading, doing well would provide defeasible, probabilistic evidence of intelligence rather than a definition of it.
Objections and replies
Objections , Raised by other thinkers and critics, as reported

The nine objections listed above are the ones Turing raised against himself. The Stanford entry adds its own doubts about several of his replies: it suggests that the reply to the mathematical objection may simply assert that human beings are free of the relevant limitation, that the reply to Lady Lovelace needs more elaboration, and that the reply to the argument from continuity would not help if real thought requires continuous states. It reports that Bringsjord and colleagues (2001) called Turing's answer to Lady Lovelace 'mysterious' at best and 'incompetent' at worst. The entry also separates two claims that critics often run together: that the game is a good test for intelligence, and that a programmed digital computer could pass it.

Replies

Turing's own replies, as the Stanford entry reports them, are brief. To the mathematical objection he answers that it is not clear that human beings are free of the limitations Gödel's theorem places on machines, while granting that the argument cannot be dismissed lightly. To the argument from consciousness he answers that the objector risks solipsism, and that he would be satisfied if we agreed that we have as much reason to think machines think as to think other people do; he offers a sample viva voce exchange as illustration. To the arguments from disabilities (the claim that a machine could never be kind, enjoy strawberries and cream, make mistakes and so on) he answers that the doubt rests largely on our limited past experience of machines. To Lady Lovelace he asks whether we ever do anything really new, and notes that machines take us by surprise. To the argument from continuity he answers that a discrete-state machine can imitate a continuous one with very small error. On informality of behaviour he notes a slide between 'ought' and 'will' in the objection's formulation, and on extra-sensory perception he proposes a telepathy-proof room. The theological and 'heads in the sand' objections receive no substantive answer in the entry's report.

Coordinate 6.17Coordinate 1.3Alan TuringJohn SearleNed BlockSources and primary textsRead the full entry

What it means

The Stanford Encyclopedia entry by Graham Oppy and David Dowe notes that the phrase is most properly used of a proposal Alan Turing made in 'Computing Machinery and Intelligence' (1950). In Turing's imitation game an interrogator questions two hidden participants, a human and a machine, by typed messages and tries to say which is which. Turing predicted that in about fifty years' time machines would play so well that an average interrogator would have no more than a 70 per cent chance of making the right identification after five minutes of questioning. The entry separates this empirical prediction from the conceptual question of what passing would show.

The question. Can the question whether machines can think be settled at all, and if so, what kind of evidence would settle it? Turing's essay asks whether the ordinary question can be replaced by one that admits of precise discussion, and whether the objections to machine thinking defeat the replacement.

Provenance. The argument originated with Alan Turing, 'Computing Machinery and Intelligence', Mind, new series, 59, no. 236 (October 1950), pp. 433–460. The Turing Digital Archive at King's College, Cambridge holds an off-print (AMT/B/19), reproduced with the permission of Oxford University Press. The label 'Turing test' is later; Turing called his proposal the imitation game.

The text's own reasoning. As the Stanford entry reports it, Turing judged the original question 'too meaningless' to deserve discussion (p. 442) and proposed instead to ask whether a digital computer could do well in the imitation game, a question that, in his eyes, admits of precise discussion. He did not claim a proof. Most of the essay answers objections, and Turing himself said he had 'no very convincing arguments of a positive nature to support my views' (p. 454, as quoted in the entry).

Structure. (COT's reconstruction; Turing does not number his steps) (1) The question 'Can machines think?', as ordinarily posed, is too unclear to be discussed profitably. (2) The question whether a digital computer could do well in the imitation game is precise and can be investigated. (3) If a machine did well in the game, we would have as much reason to think that it thinks as we have to think that other people think; this premise is drawn from Turing's reply to the argument from consciousness. (4) None of the objections Turing considers shows either that a digital computer could not do well in the game or that doing well would not count. (5) So the precise question may stand in for the original one, and the original need not be answered first.

Intended conclusion. The question whether machines can think should be approached through the question whether a machine could do well in the imitation game, and Turing expected the answer to the second question to be yes. On the Stanford entry's reading, doing well would provide defeasible, probabilistic evidence of intelligence rather than a definition of it.

On the entry's reading, the test is probabilistic, and passing it provides no more than defeasible evidence of intelligence. Turing himself considered nine objections: theological; the 'heads in the sand' objection; the mathematical objection from Gödel's theorem; the argument from consciousness; arguments from various disabilities; Lady Lovelace's objection that machines can only do what they are programmed to do; the continuity of the nervous system; the informality of behaviour; and extra-sensory perception.

Objections. The nine objections listed above are the ones Turing raised against himself. The Stanford entry adds its own doubts about several of his replies: it suggests that the reply to the mathematical objection may simply assert that human beings are free of the relevant limitation, that the reply to Lady Lovelace needs more elaboration, and that the reply to the argument from continuity would not help if real thought requires continuous states. It reports that Bringsjord and colleagues (2001) called Turing's answer to Lady Lovelace 'mysterious' at best and 'incompetent' at worst. The entry also separates two claims that critics often run together: that the game is a good test for intelligence, and that a programmed digital computer could pass it.

Replies. Turing's own replies, as the Stanford entry reports them, are brief. To the mathematical objection he answers that it is not clear that human beings are free of the limitations Gödel's theorem places on machines, while granting that the argument cannot be dismissed lightly. To the argument from consciousness he answers that the objector risks solipsism, and that he would be satisfied if we agreed that we have as much reason to think machines think as to think other people do; he offers a sample viva voce exchange as illustration. To the arguments from disabilities (the claim that a machine could never be kind, enjoy strawberries and cream, make mistakes and so on) he answers that the doubt rests largely on our limited past experience of machines. To Lady Lovelace he asks whether we ever do anything really new, and notes that machines take us by surprise. To the argument from continuity he answers that a discrete-state machine can imitate a continuous one with very small error. On informality of behaviour he notes a slide between 'ought' and 'will' in the objection's formulation, and on extra-sensory perception he proposes a telepathy-proof room. The theological and 'heads in the sand' objections receive no substantive answer in the entry's report.

Ned Block's 'Blockhead' thought experiment imagines a vast look-up table that could produce human-like replies while having no more intelligence than a toaster; the entry asks whether such a machine is even physically possible. John Searle's Chinese room argument is sometimes taken as an objection to the test, but the entry argues that it challenges Turing's expectation that digital computers will think rather than the usefulness of the test itself. The entry also reviews the Loebner Prize competitions, whose entrants were easily detected, the 2014 claims made for the chatbot 'Eugene Goostman', large language models such as GPT-3, and proposed alternatives such as the Total Turing Test and the Lovelace Test.

Key thinkers and texts. Alan Turing, 'Computing Machinery and Intelligence' (1950); Ned Block's Blockhead argument; John Searle's Chinese room argument; Copeland (2000), Piccinini (2000) and Moor (2001) for the standard reading of the game; Genova (1994), Traiger (2000) and Sterrett (2000, 2020) for readings on which the machine imitates a woman.

Current status. The essay remains the starting point for debates about machine intelligence, but there is no agreement on how to read it. The Stanford entry records a dispute over whether the machine is meant to imitate a human being or, as in Turing's opening version with a man and a woman, a woman, with Copeland, Piccinini and Moor defending the standard reading and Sterrett dissenting. It also records rival views of what passing would show: a logically sufficient condition (the reading attacked by Block's argument), a criterion, or probabilistic evidence (Moor; Bringsjord).

Related entries on this site. The Chinese room argument and the Chinese nation argument are set out in their own entries and are not repeated here; Moral status of artificial intelligence and Strong AI take up what would follow if machines did think. The book Computing Machinery and Intelligence in Book Corner gives the bibliographic details of the essay.

On this site it belongs to Coordinate 6.17 (Philosophy of Artificial Intelligence) and Coordinate 1.3 (Philosophy of Mind).

See also the glossary entries Chinese room argument, Computational theory of mind, Functionalism, Consciousness, Thought experiment and Solipsism.

The questions it belongs to

Thinkers

To read

Compare this argument

Sources and further reading

Responsibility: Coordinates of Thought Editors. Checked against the sources above on 15 November 2026. Not peer-reviewed by outside specialists. How we write and check.

Cite this page

Responsibility: Coordinates of Thought Editors. Each statement was drafted from the sources listed above and checked against them in a separate pass before publishing; this page is not peer-reviewed by outside specialists. Source-checked 15 November 2026. How to cite this site · How we write and check.

APA (7th edition)

Coordinates of Thought Editors. (2026, November 15). Turing test (Arguments). Coordinates of Thought. Retrieved [date accessed], from https://coordinatesofthought.com/arguments/turing-test/

MLA (9th edition)

“Turing test (Arguments).” Coordinates of Thought, 15 Nov. 2026, coordinatesofthought.com/arguments/turing-test/. Accessed [date accessed].

Chicago (17th edition)

Coordinates of Thought Editors. “Turing test (Arguments).” Coordinates of Thought. Source-checked November 15, 2026. https://coordinatesofthought.com/arguments/turing-test/.

BibTeX

@misc{cot-glossary-turing-test-2026,
  author       = {{Coordinates of Thought Editors}},
  title        = {Turing test (Arguments)},
  howpublished = {Coordinates of Thought},
  year         = {2026},
  month        = nov,
  url          = {https://coordinatesofthought.com/arguments/turing-test/},
  note         = {Source-checked 15 November 2026}
}

RIS

TY  - ELEC
AU  - Coordinates of Thought Editors
TI  - Turing test (Arguments)
T2  - Coordinates of Thought
PB  - Coordinates of Thought
PY  - 2026
DA  - 2026/11/15
UR  - https://coordinatesofthought.com/arguments/turing-test/
LA  - en
ER  - 

For reference managers (Zotero, Mendeley, EndNote): BibTeX file · RIS file

All arguments · The Philosophy Glossary