Computing Machinery and Intelligence

Turing argues that the question 'Can machines think?' is too vague to answer productively, and he replaces it with a behavioural test: can a machine's typed responses deceive an interrogator into thinking it is human? The paper rejects the demand for a definition of 'thinking' before any empirical investigation can begin. Turing's wager is that if a machine can pass the imitation game, the metaphysical question of whether it 'really' thinks loses its urgency. He does not claim to prove that machines think; he proposes a criterion that makes the question tractable and shifts the burden onto those who would deny machine intelligence without offering a testable alternative.

The original imitation game is a three-player party game: a man (A), a woman (B), and an interrogator (C) who communicates by typed messages and tries to identify which is the man and which is the woman. Turing then asks what happens when a machine takes the place of A. The version that became known as the 'Turing test'—a human interrogator judging between a machine and a human—was one of several formulations Turing offered, including an equivalent version in the paper and a third in a 1952 BBC broadcast . As Saul Traiger and others have noted, the standard interpretation often simplifies Turing's more nuanced original presentation, which was less a laboratory protocol than a rhetorical and philosophical device.

Turing addresses the theological objection (thinking requires a soul), the 'heads in the sand' objection (consequences too dreadful), the mathematical objection (Gödelian limits), the argument from consciousness (machines cannot feel), arguments from various disabilities (machines cannot do X), Lady Lovelace's objection (no originality), the argument from continuity in the nervous system, the argument from informality of behaviour, and extra-sensory perception . The consciousness objection is arguably the most philosophically enduring, and Turing's reply—that we cannot know other minds except through behaviour—anticipates the 'other minds' response later associated with the debate. John Searle's Chinese room argument, advanced in 1980, is a sophisticated descendant of this objection .

Turing treats ESP as a live empirical possibility in 1950 and argues that if telepathy existed, it could allow a human interrogator to identify a machine by mind-reading rather than by conversational performance. He proposes that the test be conducted in a 'telepathy-proof room' to rule this out . The inclusion of ESP is often treated as an embarrassment or a curiosity, but it reveals something important about the paper's method: Turing is willing to take seriously any empirical hypothesis that could undermine the test's validity. The gesture is consistent with his broader approach—intelligence is to be investigated empirically, not defined a priori, and even fringe possibilities deserve consideration if they bear on the test's conditions.

A common misreading treats the imitation game as Turing's proposed definition of intelligence, or as a serious engineering benchmark that AI research should aim to pass. Turing's friend Robin Gandy said the paper was intended 'not so much as a penetrating contribution to philosophy but as propaganda'—Turing wanted to shake philosophers and scientists out of the assumption that computers were mere calculating engines . The paper is rhetorical and polemical as much as it is technical. Turing's prediction that machines would pass the test by 2000 was a provocation, not a research target. Reading the paper as a literal specification for a laboratory experiment misses its philosophical function: to make the question of machine intelligence respectable by giving it a concrete, debatable form.

The reception history has been periodized in several phases. In the 1950s, the paper was often read as a neo-behaviourist proposal in philosophy and as a preliminary definition of machine intelligence in the emerging AI field. From the 1960s through the 1990s, the test became a focal point in debates about mechanism and consciousness, with Searle's Chinese room argument (1980) marking a turning point . From the 1990s onward, computer scientists and AI researchers largely rejected the test as a meaningful benchmark, while philosophers continued to defend and refine it . The Loebner Prize competitions, which attempted to operationalize the test, are now widely regarded as having demonstrated its limitations as a practical measure rather than its philosophical depth.