This is part 2 of our extended history of AI. We’re running weekly installments in this newsletter. If you missed part 1, find it here.
“I propose to consider the question, ‘Can machines think?’”
Today, it’s a question being discussed by college kids sitting around stoned in their dorm rooms; when British mathematician Alan Turing asked it in 1950, it was unprecedented.
Turing was speaking 14 years after he released a paper introducing the abstract “Turing machine,” later recognized as foundational to modern computation. His new paper – “Computing Machinery and Intelligence” – looked at the metaphysics of a modern computer.
“Thinking” was notoriously slippery to define, with philosophers having spent centuries failing to pin it down. Turing thus proposed a substitution: Forget what thinking is and focus on what thinking looks like.
He called it the “imitation game”: A human interrogator would exchange written messages with two hidden parties, one human, one machine. If the interrogator could not reliably tell which was which, the machine had passed. Not necessarily because it thinks, but because no observable difference remains.
Turing therefore changed the way we think of thinking, and helped pave the way for the concept of artificial intelligence. Can the machine hold a conversation indistinguishable from a human’s? If yes, whatever is happening inside – gears, circuits, and silicon – becomes irrelevant.
Yet there was a cost to this simplicity: Turing linked intelligence and dialogue, suggesting that linguistic fluency equals understanding. As anyone using ChatGPT today knows, chatbots can seem smart while being dumb.
But Turing’s test was less about the finishing line than the starting point – a way to structure research when no one could agree on definitions.
This came at a pivotal moment.
Computing was no longer a wartime secret in 1950, but it was still funded by states, mainly for state purposes, like cryptography, ballistics, and scientific calculation. Yet it was increasingly being asked what else computers could do: What could they automate? What jobs or parts of the economy could they alter? How far could they go?
The “Turing Test,” as it came to be called, was never meant as a prophecy. Turing didn’t claim machines would think, or that passing the test proved thought. He offered something more modest and more durable, though, which was a way to argue and discuss.
But after being published in Mind, a British philosophy journal with a small academic readership, there was no telling what would come. For most of the 1950s, the paper circulated quietly among philosophers and a handful of computer scientists. The term “Turing Test” didn’t even come to exist until years later, coined by others summarizing his idea for new audiences.
Even in its obscure journal, though, Turing shifted the context: As the 1950s progressed, a small but growing network of mathematicians, engineers, psychologists, and neurologists began converging on a shared ambition of building machines that could do more than calculate. They wanted machines that could learn, reason, play games, and prove theorems. They needed a vocabulary and a set of problems to solve.
Turing’s paper gave them tools and thoughts to organize around. It was the clearest articulation of the underlying question: What would it mean for a machine to be intelligent? Whether you agreed with his answer or not, you now had to respond to it.
Yet Turing would never live to see his impact.
Despite being a successful codebreaker who helped the British crack German communications during World War II, Turing was prosecuted for “gross indecency,” the British criminal charge for homosexuality at the time. He had reported a burglary to the police and, during the investigation, admitted to a sexual relationship with a man. Rather than prison, he accepted chemical castration: A year of estrogen injections meant to suppress his libido.
In June 1954, he was found dead in his home, officially from suicide by cyanide poisoning. While an apple laced with cyanide was found by his bedside, some still dispute the ruling. Regardless, one of the world’s brightest minds was dead at 41.
Yet his ideas would live on: By the mid-1950s, the conversation he had helped stimulate had grown large enough to need a room. In the summer of 1956, a small group of researchers gathered at Dartmouth College in New Hampshire to formalize what they were doing – and to give it a name.
They called it “artificial intelligence.”
Editor’s Note
Thanks for reading. Are you enjoying our history of AI series? Let us know by emailing us here.
And, as always, you can find our most recent deep-dives below:
See you tomorrow.
—Max and Max
Turing was officially apologised to in 2009 and pardoned in 2013. Too little, too late…but still.
We can only hope that the AI trying to help humans is smarter than the AI that will try to destroy us!