Artificial intelligence as a science became possible only with the invention of programmable electronic computers in the 1940s. One person, more than any other, laid its intellectual foundations: the English mathematician Alan Turing.

The universal machine

In 1936, at the age of twenty-four, Turing published a paper on a problem in mathematical logic. To solve it, he imagined a simple abstract machine that reads and writes symbols on an endless tape, following a table of rules. He showed that a single universal machine could imitate any other such machine, if given a description of it. This idea, now called the Turing machine, defined what it means for a problem to be computable, and it is the theoretical basis of every modern computer, which can run any program rather than being built for one task.

War and codebreaking

During the Second World War, Turing worked at Bletchley Park, Britain's secret codebreaking center. He played a leading role in breaking the German Enigma cipher, designing electromechanical machines called bombes that searched for the daily settings. The intelligence gained helped the Allies, especially in the Battle of the Atlantic. Elsewhere, engineers built some of the first electronic computers: Colossus at Bletchley to attack another German cipher, and in the United States, ENIAC, completed in 1945, a room-sized machine with about eighteen thousand vacuum tubes.

Neurons as logic

In 1943, the neuroscientist Warren McCulloch and the logician Walter Pitts published a paper proposing a simplified mathematical model of a neuron: a unit that receives inputs, adds them up, and fires if the total passes a threshold. They showed that networks of such units could compute logical functions. This was the seed of the neural network approach to AI, inspired by the brain. In 1948 Claude Shannon published his theory of information, and the mathematician Norbert Wiener popularized cybernetics, the study of control and communication in animals and machines.

Can machines think?

In 1950, Turing published “Computing Machinery and Intelligence” in the philosophy journal Mind. He began with the question “Can machines think?” but argued that it was too vague to answer directly. He proposed replacing it with a test he called the imitation game, now known as the Turing test. A human interrogator holds text conversations with a hidden human and a hidden machine and tries to tell which is which. If the interrogator cannot reliably distinguish them, Turing suggested, we would have as much reason to say the machine thinks as we have for other people.

Turing predicted that in about fifty years, computers would play the game well enough that an average interrogator would have no more than a seventy percent chance of identifying them correctly after five minutes of questioning. He also answered objections, including Lady Lovelace's: he argued that machines could surprise their makers, and that a machine might learn, like a child, rather than being programmed with everything in advance. He called this idea a “child machine” that could be educated, an early vision of machine learning.

Turing's fate

In 1952, Turing was prosecuted for homosexual acts, then a crime in Britain. To avoid prison he accepted chemical treatment with hormones. He died in 1954, at forty-one, from cyanide poisoning, ruled a suicide. Decades later, in 2009, the British prime minister issued a public apology, and in 2013 Turing received a royal pardon. Since 2021 his portrait appears on the Bank of England's fifty-pound note. His story is a reminder that a society can lose a great mind through prejudice.

Dartmouth, 1956

The field got its name in a proposal written in 1955 by the American computer scientist John McCarthy, with Marvin Minsky, Nathaniel Rochester and Claude Shannon. They proposed a summer research project at Dartmouth College in New Hampshire on “artificial intelligence,” a term McCarthy coined. The proposal expressed the bold conjecture that every aspect of learning or any other feature of intelligence could in principle be described so precisely that a machine could be made to simulate it. The 1956 workshop brought together the people who would lead the field for decades. It is usually counted as the birth of AI as a discipline.

Worked example: does the Turing test work?

Weigh the strengths and weaknesses of Turing's test, as later thinkers did.

• Strength: it is practical. Instead of arguing endlessly about the definition of thinking, it offers an observable test.
• Strength: it is fair. We judge other people's minds by their behavior; the test judges machines the same way.
• Weakness: it can be fooled. Simple programs have fooled people by imitating human quirks, changing the subject or pretending to be a child or a non-native speaker.
• Weakness: imitation is not understanding. In 1980, the philosopher John Searle described the Chinese Room: a person who follows rules to produce correct Chinese answers without understanding Chinese. Searle argued that computers manipulate symbols in the same way, without understanding.

By the 2020s, large language models could hold conversations that many people found hard to distinguish from a human's, yet experts still disagreed about whether they understood anything. Turing's test turned out to measure how convincing a machine is, which is important, but not to settle the deeper question he set aside.