Chapter 4: The Turing Prophecy

In the autumn of 1948, a man walks to work through Manchester streets still scarred by German bombs. Almost no one he passes knows what he did in the war, how his work at Bletchley Park helped break the Nazi codes, how his ideas contributed to the machines that cracked Enigma. That will remain secret for decades. The man is Alan Turing, and at the university a new machine awaits him: one of the first electronic computers ever built.

But Turing is not thinking about calculation. He is thinking about something stranger. He is thinking about whether the machine might one day think.

This is the peculiar thing about Turing. He invents the theoretical framework for computation itself—the abstract "Turing machine" that defines what any computer can and cannot do—and then immediately asks whether computation might produce mind. He builds code-breaking devices that help win a war, and wonders if such devices could write poetry. He is not interested in faster arithmetic. He is interested in the nature of intelligence.

By 1950, he has been arguing with distinguished colleagues for years. They do not believe him.

Geoffrey Jefferson, the eminent neurosurgeon, had delivered a lecture the previous June that drew a line Turing found indefensible. "Not until a machine can write a sonnet or compose a concerto because of thoughts and emotions felt," Jefferson proclaimed, "and not by the chance fall of symbols, could we agree that machine equals brain." Douglas Hartree, the physicist, made similar objections. Michael Polanyi, the chemist-turned-philosopher, pressed Turing on the impossibility of formalizing tacit knowledge, the things we know how to do without being able to explain.

These were serious men making serious arguments. They believed consciousness was the prerequisite for genuine intelligence. Without the felt quality of experience, any output was mere mimicry. The machine might produce symbols that resembled thought, but there would be nobody home.

Turing responded with a paper.

He wrote it quickly, with evident enjoyment. His student Robin Gandy later observed that it was "intended not so much as a penetrating contribution to philosophy but as propaganda." Turing wanted to change how people thought about thinking machines, and he knew he needed something memorable.

The paper appeared in the journal Mind in October 1950. It was called "Computing Machinery and Intelligence," and it began with a game.

"I propose to consider the question, 'Can machines think?'"

The famous opening conceals a rhetorical sleight of hand. Turing immediately acknowledges that the question is almost meaningless as stated, the words "machine" and "think" too imprecise to permit rigorous analysis. So he replaces the question with a test.

The test involves three parties: a human interrogator, a human respondent, and a machine. The interrogator communicates with both through text alone, asking whatever questions might reveal which is which. If the interrogator cannot reliably distinguish the machine from the human, Turing argues, we should say the machine thinks.

This is the Imitation Game, now known as the Turing Test. Its brilliance lies in what it sidesteps. Jefferson had demanded proof of inner experience, but how do we prove inner experience in anyone? We assume other people are conscious because they behave as if they are, because their words and actions cohere with our own felt sense of what consciousness produces. We do not open their skulls and inspect their qualia.

Turing forces the critics into an uncomfortable corner. Unless you embrace solipsism and deny that any mind but your own truly thinks, you cannot in principle demand from machines what you cannot demand from humans.

"I am sure that Professor Jefferson does not wish to adopt the extreme and solipsist point of view," Turing wrote, with the air of a man laying a trap.

The test has limits Turing himself acknowledged.

It does not ask whether the machine is conscious—only whether its behavior is indistinguishable from that of a conscious being. A perfect mimic could pass the test while being, in some deep sense, hollow. The Chinese Room argument, which John Searle formulated three decades later, attacks precisely this point: following rules for manipulating symbols is not the same as understanding them, no matter how convincing the output.

Turing knew this. "I do not wish to give the impression that I think there is no mystery about consciousness," he wrote. "But I do not think these mysteries necessarily need to be solved before we can answer the question with which we are concerned."

This is the crucial methodological move. Turing is not claiming machines are conscious. He is saying something more modest and more revolutionary: we can make progress on machine intelligence without first solving the mind-body problem. We can build systems that behave intelligently and defer the question of whether there is "something it is like" to be such a system.

His critics had insisted on inner experience as the criterion. Turing shifted to observable behavior, not because he denied the importance of experience, but because experience cannot be observed from outside. The test that bears his name is an operational definition, a way forward that doesn't require settling metaphysics first.

The limits of the test are real. But so is its pragmatic power.

In that same paper, Turing catalogued nine objections to thinking machines and responded to each. The Religious Objection: thinking is a function of the immortal soul. (Turing, gently: "I am unable to accept any part of this.") The "Heads in the Sand" Objection: the consequences of machines thinking would be too dreadful to contemplate. (Turing: this is mere sentiment, not argument.) Lady Lovelace's Objection: machines cannot originate anything, only do what we program them to do. (Turing: surprises emerge from rules whose consequences we cannot foresee.)

But the most revealing section comes near the end, where Turing considers how such a machine might actually be built.

His suggestion sounds radical for 1950: don't try to program adult intelligence directly. Instead, program a child.

"Why not rather try to produce one which simulates the child's? If this were then subjected to an appropriate course of education one would obtain the adult brain."

The child machine would need mechanisms for reward and punishment, random elements to enable exploration, ways to form associations between stimuli and responses. It would learn from experience rather than having its knowledge pre-specified. "We may hope that machines will eventually compete with men in all purely intellectual fields."

This is machine learning, outlined decades before the hardware existed to attempt it seriously. Turing saw that the path to intelligence ran through development and education, not through programming complete knowledge. Build simple systems that can learn, rather than complex systems that already know.

The insight would wait a generation to be rediscovered. When neural networks finally succeeded, they succeeded by following Turing's prescription: start simple, learn from examples, let complexity emerge.

In his final years, Turing turned to biology.

The question that gripped him was morphogenesis: how does a spherical embryo become a non-spherical organism? How do zebra stripes emerge, or leopard spots? The developing organism starts uniform—the same cells everywhere—and somehow organizes itself into structured complexity.

Turing proposed that chemical substances he called "morphogens," reacting and diffusing through tissue, could account for this emergence. His mathematical model showed how random fluctuations could drive pattern formation from initial uniformity. Start with nothing special, and specialness emerges through the dynamics of the system.

"The Chemical Basis of Morphogenesis" was published in 1952. Decades later, scientists validated his predictions, showing that leopard spots and zebra stripes and the branching of blood vessels all followed the patterns his equations described.

Here is the thread connecting all of Turing's work: how does complexity emerge from simplicity? In computation, he asked how universal machines could arise from simple operations. In intelligence, he asked how mind-like behavior could emerge from mechanical processes. In biology, he asked how organized structure could emerge from chemical gradients. The question was always the same, wearing different masks.

He did not live to see where these questions led.

In 1952, the same year the morphogenesis paper appeared, Turing was prosecuted for "gross indecency," the legal term for his relationship with another man. His wartime service, his fundamental contributions to computation, his insights into the nature of mind—none of this mattered to a society that criminalized his love.

He was given a choice: prison or chemical castration. He chose the hormones. The treatment caused physical changes he found humiliating. His security clearance was revoked; the man who had helped crack Enigma was now considered a security risk because of who he loved.

On June 7, 1954, his housekeeper found him dead. An apple beside his bed, partially eaten, was suspected to contain cyanide. He was 41 years old.

The prophet of machine intelligence was destroyed by human cruelty—by a society that could not tolerate difference in a man whose entire intellectual life was devoted to understanding what difference meant.

There is no way to know what Turing would have contributed had he lived. He was just beginning the morphogenesis work. He had ideas about neural networks that he never fully developed. He was thinking about the relationship between randomness and creativity, between simplicity and emergence. The patterns he saw in biology and computation might have merged into something new.

What we have instead is the test, the prophecy, and the questions.

Can machines think? Turing's answer: try to tell the difference, and see if you can. Don't demand proofs of inner experience that you cannot provide for yourself. Judge by what the system does, not by metaphysical prejudice about what it must be.

Build child machines that learn. Start simple. Let complexity emerge.

And never forget: the question of what machines can become is inseparable from the question of what we allow ourselves to be. Turing was punished for a love his society refused to recognize. His machines were dismissed because his critics refused to imagine what computation might become. Both refusals arose from the same failure—the inability to recognize new patterns as legitimate.

Two years after Turing's death, a group of researchers gathered at Dartmouth College to found a new field. They called it artificial intelligence. They built on Turing's foundations, often without knowing quite how much they owed him.

The prophecy began to come true. But the prophet was already gone.