The Turing Test's Blind Spot

Turing's biggest proof was its own limitation: computers can do more than imitate men.

Turing's genius was to compare machines with humans. Because of him, machines now stand on their own.

[optional image description]
Wikimedia Commons

To any outsider it was an unlikely place, but the modern era of computers began on May 28, 1936, when the editors of the Proceedings of the London Mathematical Society received a paper with the rather cumbersome title, "On Computable Numbers, With an Application to the Entscheidungsproblem" by Alan M. Turing.

The Entscheidungsproblem, or in English "the decision problem," was a challenge put forth by the German mathematician David Hilbert in an esoteric field of research on the foundations of mathematics.

Little did anyone know at the time that this obscure 35-page paper from a 23-year-old would build a road from logic and mathematics to cryptography, artificial intelligence, and an inquiry into the nature of our minds.

A celebration of the life and work of the pioneering computer scientist
See full coverage

Turing's 1936 paper on computable numbers hit that rare bull's eye where philosophy and discovery overlap. The mathematician Alonzo Church at Princeton came to the same findings just months before Turing. But unlike Church, who used the standard abstractions of pure mathematics in his argument, Turing wrote of machines, algorithms, ink, paper tape, and computation. (Before Turing, a "computer" referred not to a machine, but to a human being who calculated with paper and pencil.)

This unique blend of philosophical speculation, eccentricity, and logical rigor was Turing's signature across his astonishing career. Later on, after the war, after he had snapped the German Engima cipher (while wearing pajamas), he took it upon himself to defend thinking machines. We would know machines could think, he claimed, when we couldn't tell if we were talking to a computer or a man, provided a wall or a door prevented us from seeing which was which.

The Turing Test asks the deepest questions about what makes human intelligence, if at all, unique. What's odd is that so odd a man would seek to vindicate artificial intelligence by showing how it could imitate men. The truth is that it's capable of so much more.

From our fingerprints to our minds, we tend to believe that humans are very different from each other. Take a random set of talented people -- Nietzsche, David Deutsch, Michael Phelps, Ada Lovelace, Louis C.K., Charles Dickens, you and your friends -- and it is striking how different their geniuses are. Expand that set further to include all of mankind's differing abilities, interests, personalities, and talents. The task's scope boggles the mind.

Now think of computers, and this vast picture narrows quickly. From the bit to the interface, most people tend to see computers as all the same: a black box executing programs with the push of a button. Even the range of their physical shape seems extraordinarily small. Outside of humanoid robots in science fiction, most people struggle to imagine anything beyond desktops, laptops, phones, and tablets.

But while most people think that humans differ greatly and that computers are all pretty much the same, the truth is very different. In fact, the range of potential applications of computer intelligence is much larger than the range of human intelligence.

Cars are on the verge of becoming driverless. Market-making high-frequency traders have automated the trading pit. Soon, computer vision may improve medical diagnostics beyond the ken of what any human could detect. This is only the beginning. Indeed, the arrow of technological progress may soon hit an inflection point of rapid acceleration leading to a world of intelligent tools and entities, both utterly different and unrecognizable from any intelligence we've ever worked with before.

The origins of this Cambrian explosion in intelligence began with Turing's proofs and thought experiments. One hundred years after his birth it is fitting and right that we should honor his life and accomplishments, but there is also a lesson.

Turing tried to show that machines could act like men, but the tragic irony of his life is that it ended with men expecting him to act like a reprogrammable machine. After his conviction for "gross indecency," the British government forced Turing to take estrogen to "cure" his homosexuality. Unable to bear this burden, he killed himself on June 7, 1954.

Not every human thinks alike; but the little known secret is that not every machine does, either. In order to think different, we have to act different, whether we are men or machines.