In October 1950, the British journal Mind published a paper by Alan Turing titled “Computing Machinery and Intelligence.” It began with a sentence that would become one of the most quoted in the history of science: “I propose to consider the question, ‘Can machines think?'”
The paper was unlike anything that had appeared in a philosophy journal before. Its author was not a philosopher but a mathematician and logician who had, during World War II, helped break the German Enigma cipher at Bletchley Park and who had, in 1936, defined the theoretical foundations of computation with his concept of the universal Turing machine. Now, just five years after the first electronic computers had been built, Turing was asking whether those machines might eventually be capable of thought.
The paper is remarkable not only for the question it posed but for the quality of the answers it anticipated. Turing considered and responded to nine separate objections to machine intelligence, addressing arguments from theology, mathematics, consciousness, and common sense with a clarity and intellectual honesty that remain impressive seventy-five years later. Nearly every debate about artificial intelligence that has occurred since 1950 was anticipated, in some form, in Turing’s paper.
The Imitation Game
Turing recognized that the question “Can machines think?” is poorly defined. The words “machine” and “think” are too vague to permit a rigorous answer. Rather than attempt to define thinking, he proposed replacing the question with a practical test.
The test, which Turing called the imitation game (later known as the Turing test), works as follows. A human interrogator communicates by text with two respondents: one human and one machine. The interrogator does not know which is which. Through a series of questions and answers, the interrogator attempts to determine which respondent is human and which is the machine. If the machine can fool the interrogator consistently, Turing argued, we should consider the machine to be thinking, or at least we should not deny it the capacity for thought on any principled grounds.
Turing was careful about what the test does and does not establish. He did not claim that passing the test proves a machine is conscious. He claimed that if a machine’s behavior is indistinguishable from a human’s, the question of whether it “really” thinks becomes meaningless, just as the question of whether another person “really” thinks (as opposed to merely behaving as if they think) is ultimately unanswerable.
The Nine Objections
The bulk of the paper is devoted to considering and responding to objections. Turing listed nine, each representing a different reason why someone might deny that machines can think.
The theological objection: thinking is a function of the soul, which God has given only to humans. Turing noted that theological arguments about the limits of God’s power are unconvincing and that this objection would have been equally applicable to any sufficiently novel technology.
The “heads in the sand” objection: the consequences of machines thinking would be too terrible, so it cannot be true. Turing dismissed this as wishful thinking.
The mathematical objection: Gödel’s incompleteness theorems show that there are mathematical truths that no formal system can prove. Since computers are formal systems, there are things they cannot do that humans can. Turing’s response was subtle: he noted that humans are also limited in their mathematical abilities and that there is no evidence that humans can do things that are provably impossible for machines.
The consciousness objection: a machine cannot truly think because it lacks subjective experience. Turing acknowledged the force of this objection but pointed out that we have no way of knowing whether other humans have subjective experience either. The only evidence we have for other minds is behavior, and the imitation game tests behavior.
Arguments from various disabilities: machines cannot appreciate beauty, fall in love, learn from experience, or do anything genuinely new. Turing argued that these are all claims about specific capabilities and that future machines might possess them. The fact that current machines lack these abilities proves nothing about machines in principle.
Lady Lovelace’s objection: machines can only do what they are programmed to do; they cannot originate anything. Turing’s response was one of his most prescient: he noted that a machine’s output can surprise its programmer, just as a student’s work can surprise a teacher. The fact that a machine follows rules does not mean its behavior is predictable or unoriginal.
The continuity argument: the nervous system is continuous (analog) while computers are discrete (digital). Turing showed that discrete machines can simulate continuous processes to any desired accuracy, making this distinction irrelevant in practice.
The informality of behavior: human behavior cannot be captured by rules; therefore, machines (which follow rules) cannot reproduce human behavior. Turing questioned the premise, arguing that there is no evidence that human behavior is not, at some level, rule-governed.
Extrasensory perception: Turing took this objection seriously (ESP was considered more credible in 1950 than it is today) and suggested that a telepathy-proof room might be needed for a fair test. This is generally regarded as the least successful section of the paper.
Learning Machines
The final section of the paper contains Turing’s most forward-looking ideas. He argued that rather than trying to program a machine to simulate an adult human mind directly, it might be more practical to build a machine that simulates a child’s mind and then educate it. The machine would learn from experience, modifying its own behavior in response to rewards and punishments, much as a child does.
This idea anticipates the central approach of modern machine learning by decades. Contemporary neural networks learn by adjusting their internal parameters in response to training data, a process that is, in broad outline, exactly what Turing described. The large language models that power modern AI systems (such as GPT and its successors) are trained on enormous datasets and modify their behavior through a learning process, not through explicit programming of rules.
Turing also predicted that a thinking machine would make mistakes, that it would sometimes behave unpredictably, and that its internal processes might be too complex for its programmers to fully understand. All of these predictions have proved accurate for modern AI systems.
The Paper’s Reception
The paper was published in Mind, one of the oldest and most respected philosophy journals in the English-speaking world. Its reception was mixed. Philosophers debated the adequacy of the imitation game as a test for thinking. Scientists were more interested in the practical suggestions about learning machines. The general public, to the extent it was aware of the paper at all, found the idea of thinking machines either exciting or alarming.
The term “Turing test” was not coined until later (Turing himself called it the imitation game). The test became the most famous thought experiment in the philosophy of mind and the unofficial benchmark for artificial intelligence research, even though no AI system passed a rigorous version of it until the 2010s (and even then, the results were disputed).
The paper’s influence grew over the following decades as the field of artificial intelligence developed. Every major debate in AI, from the “Chinese Room” argument of John Searle (1980) to the current debates about large language models and consciousness, takes Turing’s paper as its starting point.
Turing’s Legacy
Turing did not live to see the field he envisioned. He died on June 7, 1954, at the age of forty-one, from cyanide poisoning. His death was ruled a suicide, though some biographers have argued it may have been accidental. Turing had been prosecuted in 1952 for homosexuality (then a criminal offense in Britain) and subjected to chemical castration. The treatment left him physically and emotionally devastated.
The injustice of Turing’s treatment has been widely acknowledged. He received a posthumous royal pardon in 2013, and the “Alan Turing law” (2017) retroactively pardoned thousands of men convicted under the same statutes. His face appears on the Bank of England’s fifty-pound note, and the Turing Award (the “Nobel Prize of computing”) is named in his honor.
But Turing’s greatest legacy is intellectual. His 1936 paper on computability defined what a computer is. His wartime work at Bletchley Park demonstrated what computers could do. And his 1950 paper asked what computers might become. The three papers, taken together, are the intellectual foundation of the digital age.
The Documents
Turing’s wartime work at Bletchley Park, which shaped his thinking about machine intelligence, is documented in Kronecker Wallis’s edition of Turing’s Treatise on the Enigma. Known as “The Prof’s Book” among his Bletchley Park colleagues, the manuscript describes the cryptanalytic methods that Turing developed to break the Enigma cipher. The edition preserves Turing’s handwritten corrections and annotations, offering a direct view of his working methods.
The stored-program computer that made Turing’s vision of artificial intelligence physically possible is described in the EDVAC Report by John von Neumann. Printed on blue Fabriano paper in monospace type, the report is the architectural blueprint for the machines on which every AI system runs. Together, Turing’s theoretical vision and von Neumann’s practical architecture created the foundations on which the entire field of artificial intelligence has been built.
The Question That Will Not Go Away
Three quarters of a century after Turing asked “Can machines think?”, the question remains unanswered. Modern AI systems can generate text, recognize images, translate languages, and defeat world champions at complex games. Whether they think is still debated. Turing’s paper does not settle the question, but it frames it with a precision and honesty that no subsequent treatment has surpassed.
The most remarkable thing about the paper may be its tone. Turing writes without dogmatism or defensiveness. He acknowledges the strength of the objections he considers. He makes predictions modestly, noting that he might be wrong. And he ends with a statement that captures the spirit of scientific inquiry itself: “We can only see a short distance ahead, but we can see plenty there that needs to be done.” Seventy-five years later, there is still plenty that needs to be done. Turing saw it first.
No products in the cart. 
