1936: Turing’s paper on computable numbers is read

On 12 November 1936 the London Mathematical Society formally read a paper by the 24-year-old Cambridge mathematician Alan Turing. Entitled “On Computable Numbers, with an Application to the Entscheidungsproblem,” it remains one of the founding documents of computer science and, by extension, of artificial intelligence.

Turing set out to answer David Hilbert’s Entscheidungsproblem: does there exist a definite method that can decide, for any mathematical statement, whether it is true or false? To make the question precise he invented an imaginary device—now called a Turing machine—that reads and writes symbols on an endless tape according to a finite table of instructions. He then defined the computable numbers as those that such a machine can produce. Most decisively, he described a single “universal” machine that could simulate any other Turing machine once it was given that machine’s description on its tape. In a few elegant pages he had both proved that the Entscheidungsproblem is unsolvable and supplied the abstract blueprint for the stored-program computer.

The paper’s immediate audience was logicians, yet its ideas travelled quickly into engineering. During the Second World War Turing helped turn the abstract machines into concrete code-breaking devices at Bletchley Park. After the war the universal-machine concept guided the design of the ACE, EDVAC and other early electronic computers. By the time the term “artificial intelligence” was coined at Dartmouth in 1956, researchers already took for granted that any intellectual procedure that could be spelled out as a finite set of rules could, in principle, be executed by a machine of Turing’s kind.

The same paper also drew a permanent boundary. Turing showed that certain well-defined questions are undecidable: no algorithm can answer them in every case. That limit still shapes modern AI. When today’s systems confront problems of verification, alignment or open-ended reasoning, they are negotiating the frontier first mapped in 1936.

Every large language model, every neural network trained by gradient descent, every robot controller running on silicon ultimately rests on digital computers whose theoretical possibility Turing established that November day. The universal machine he described has become the invisible infrastructure of contemporary intelligence—artificial and otherwise.