In the summer of 1936, a twenty-three-year-old Cambridge mathematician named Alan Turing published a paper that would reshape the world. He was not trying to build an intelligent machine. He was trying to solve an abstract problem in mathematical logic. But in doing so, he invented the theoretical foundation for every computer — and every artificial intelligence system — that would ever exist.
This chapter follows Turing from that foundational paper through the code-breaking triumph at Bletchley Park to his direct confrontation with the question that defines AI: Can a machine think?
The Impossible Problem
The problem Turing tackled was called the Entscheidungsproblem — German for "decision problem." Posed by the mathematician David Hilbert in 1928, it asked whether there exists a mechanical procedure that can determine, for any mathematical statement, whether that statement is true or false.
Hilbert believed the answer was yes. He thought mathematics was complete and consistent, and that a systematic method could settle any mathematical question. If he was right, mathematics would be — in principle — fully automatable.
Turing proved him wrong, and the way he proved it changed everything.
To answer the question, Turing first had to define precisely what a "mechanical procedure" was. There were no computers in 1936 — the word still referred to humans doing calculations by hand. So Turing invented an imaginary device: an abstract machine that could read and write symbols on an infinitely long tape, following a set of rules.
This imaginary device — now called a Turing machine — could be configured to perform any computation that could be precisely described. Turing showed that there was a "universal" version of this machine: a single device that could simulate any other Turing machine, given the right instructions on its tape.
The universal Turing machine was the theoretical blueprint for the general-purpose computer. It showed that a single machine, given the right program, could perform any computation. You did not need to build a different machine for each task. You needed one machine and many programs.
Turing then used this framework to prove that some problems are fundamentally unsolvable — no mechanical procedure can solve them, no matter how much time or resources you have. This was a landmark result in mathematics, but its practical implications were even more significant. Turing had defined computation itself, and in doing so, he had drawn the first map of what machines could and could not do.
Bletchley Park and the First Thinking Machines
Three years after publishing his paper, Turing found himself at Bletchley Park, a Victorian estate in the English countryside that had been requisitioned as Britain's code-breaking headquarters. Nazi Germany was using the Enigma machine to encrypt military communications, and the Allies needed to break those codes to win the war.
The Enigma machine produced encryptions of staggering complexity. The number of possible settings was approximately 159 million million million — a number so large that trying every combination by hand was impossible. The Germans believed Enigma was unbreakable.
Turing and his colleagues proved them wrong. Building on earlier work by Polish mathematicians, Turing designed the Bombe, an electromechanical device that could rapidly test Enigma settings by exploiting logical contradictions. If a proposed setting led to an impossible result — a letter encrypting to itself, which Enigma could not do — the Bombe eliminated it and moved on.
The Bombe was not a computer in the modern sense. It was a special-purpose machine designed for one task. But it embodied a principle that would become central to AI: rather than exhaustively searching every possibility, use logical reasoning to eliminate impossible options and narrow the search to manageable size.
By 1942, Bletchley Park was breaking thousands of Enigma messages per day. Historians estimate that this intelligence shortened the war by at least two years and saved millions of lives.
The experience at Bletchley Park transformed Turing's thinking. He had seen machines perform tasks that seemed to require intelligence — finding patterns, making logical deductions, solving problems that no human could solve by brute force. The theoretical question of whether machines could think was no longer abstract. Turing had watched machines do things that, if a human had done them, would be called intelligent.
The ACE and the Manchester Machine
After the war, Turing threw himself into building actual computers. In 1945, he designed the Automatic Computing Engine (ACE) for the National Physical Laboratory. His design was detailed and ambitious — a stored-program computer that could modify its own instructions as it ran.
Bureaucratic delays frustrated Turing, and a simplified version of the ACE was not completed until 1950, by which time other teams had already built working computers. The Manchester Baby, built at the University of Manchester in 1948, was the first electronic stored-program computer to run a program. Turing joined the Manchester computing laboratory and began using the machine to explore ideas that fascinated him — including the possibility of machine intelligence.
At Manchester, Turing wrote some of the earliest computer programs, including a chess-playing program in 1950. The computer was too slow to actually run the program, so Turing simulated it by hand, acting as the CPU and executing each instruction himself. He played a game against a colleague using this method. The program lost, but it played a recognizable game of chess — the first time a computer algorithm had done so.
"Can a Machine Think?"
In October 1950, Turing published what may be the most influential paper in the history of artificial intelligence: "Computing Machinery and Intelligence," published in the journal Mind.
The paper opens with a deceptively simple sentence: "I propose to consider the question, 'Can machines think?'"
Turing immediately recognized that the question was tangled in definitional problems. What does "think" mean? What counts as a "machine"? Rather than getting lost in philosophical debates about consciousness and understanding, Turing proposed replacing the question with something concrete and testable.
He described what he called the "imitation game," now known as the Turing Test. A human interrogator communicates via text with two hidden entities — one human, one machine. The interrogator asks questions and tries to determine which is the machine. If the machine can fool the interrogator a significant fraction of the time, Turing argued, we should accept that it can think — or at least that the question "Can machines think?" has become meaningless.
The Turing Test was brilliant in its simplicity. It sidestepped unanswerable metaphysical questions and replaced them with an empirical test. It also, intentionally or not, established a framework that would shape AI research for decades: intelligence is as intelligence does. If a machine behaves intelligently, the internal mechanism does not matter.
Answering the Objections
The most remarkable section of Turing's 1950 paper is his systematic demolition of objections to machine intelligence. He anticipated nearly every argument that would be made against AI over the next seventy-five years and responded to each one.
The theological objection — that thinking is a function of the soul, which God gave only to humans — Turing dismissed by noting that omnipotence presumably includes the power to grant souls to machines if God chose to do so.
Lady Lovelace's objection — that machines can only do what they are told — Turing challenged directly. He pointed out that machines could surprise their creators, and that the claim "machines cannot originate anything" might reflect the limits of human imagination rather than the limits of machines.
The mathematical objection — based on Gödel's incompleteness theorems, which showed that any sufficiently powerful formal system has statements it cannot prove — Turing noted that humans are subject to the same limitations. We cannot solve every problem either.
The consciousness objection — that a machine cannot truly think because it has no subjective experience — Turing treated as a problem of solipsism. We cannot prove that other humans are conscious either; we infer it from their behavior. If a machine behaves as if it is conscious, the same inference applies.
Turing also introduced an idea that would prove prophetic: machine learning. He suggested that rather than trying to program adult intelligence directly, it might be better to build a simple machine and then educate it, much as you would educate a child. "We can only see a short distance ahead," he wrote, "but we can see plenty there that needs to be done."
Turing's Prediction
Near the end of the paper, Turing made a specific prediction: "I believe that in about fifty years' time it will be possible to programme computers, with a storage capacity of about 10⁹, to make them play the imitation game so well that an average interrogator will not have more than 70 percent chance of making the right identification after five minutes of questioning."
This prediction — that by the year 2000, computers would be able to fool 30 percent of human judges in a five-minute conversation — proved too optimistic by roughly two decades. But the fact that language models can now engage in conversations that many humans find indistinguishable from talking to a person suggests that Turing's vision was correct in its direction, if not its timeline.
The Tragedy
Alan Turing's story does not have a happy ending.
In 1952, Turing was prosecuted for homosexuality, which was a criminal offense in Britain at the time. He was convicted and given a choice between imprisonment and chemical castration through hormone injections. He chose the injections. His security clearance was revoked, and he was barred from continuing his code-breaking work.
On June 7, 1954, Alan Turing was found dead in his home from cyanide poisoning. He was forty-one years old. A half-eaten apple was found beside him. The inquest ruled his death a suicide, though some have questioned this conclusion.
Turing's contributions were largely classified and unrecognized for decades after his death. The full scope of the Bletchley Park operation was not revealed until the 1970s. In 2009, the British government issued a formal apology for Turing's treatment. In 2013, Queen Elizabeth II granted him a posthumous royal pardon. In 2021, Turing's face appeared on the Bank of England's fifty-pound note.
The field Turing helped create — artificial intelligence — was formally born two years after his death, at a summer workshop in a small New Hampshire town. We turn to that story in the next chapter.
Turing's Legacy in Today's AI
Turing's influence on modern AI is difficult to overstate. The Turing Test remains the most famous benchmark for machine intelligence, even though most researchers now consider it an imperfect measure. His concept of the universal machine — one device that can simulate any computation — is the foundation of every computer, every smartphone, and every AI system in existence.
But perhaps Turing's most enduring contribution was his willingness to take the question seriously. Before Turing, the idea of machine intelligence was the province of science fiction. After Turing, it was a research program. He showed that the question "Can machines think?" was not absurd — it was scientific, and it could be investigated with rigor.
Every language model that generates fluent text, every AI system that reasons through a complex problem, every autonomous agent that plans and acts in the world owes something to a young mathematician who, in the darkest days of a world war, glimpsed the possibility of thinking machines and spent the rest of his short life trying to make them real.