If a Machine Can Win at Chess, What Good Are We?

Some days ago in New York City, Garry Kasparov played what he called the first fair chess match between a man and a machine. It ended in a draw and was, as Kasparov saw it, a way of putting to rights the injustice he had felt during the famous chess match he played in 1997, as world champion, against the IBM computer Deep Blue. Kasparov lost then, a result he claimed had come about only through human intervention in the machine.

When I met him in 1999, Kasparov’s rage appeared not to have subsided. “Machine cannot beat me!” he shouted when I inquired about his defeat, and it was striking that the machine had caused its human opponent a form of torment to which it could never succumb. However well it simulated human thought, Deep Blue had no anxieties, no remorse. Kasparov’s very humanity was his weakness; he was beaten, in a sense, not by the cool computer but by the heat of his own mind.

Deep Blue was not the first contraption to attempt to replicate the calculations of the human mind. A long time earlier, in 1769, an “automaton chess player” was built for the empress of Austria by a Hungarian civil servant named Wolfgang von Kempelen. It was made up of a wooden figure dressed in Turkish costume and seated behind a large chest, on top of which was a chessboard. There were a number of doors and drawers in the cabinet, all of which were opened when the automaton was exhibited before the public in order to prove that there was no hidden human agent. Once its insides were revealed, a member of the audience would be asked to play against the automaton, and, more often than not, the automaton would win.

Von Kempelen’s machine demolished a number of eminent opponents -- Napoleon, Benjamin Franklin, Catherine the Great -- and several thinkers, including Edgar Allan Poe, sought to unveil its secret. From its conception to its demise in a Philadelphia fire in 1854, “the Turk,” as it was known, triggered equal amounts of anxiety and awe. Women crossed themselves on entering the exhibition room; some were reported to have fainted. When the 18th century chess master Philidor played against the automaton, he told his son that no other game had ever left him so exhausted.

The wonder and fury inspired by the Turk was similar to that felt by Kasparov when he first took on Deep Blue. Kasparov told reporters that he had to rise to the challenge “in order to protect the human race.” After all, if a computer could play chess, if it could think, what would be the purpose of human beings? What would be the difference between a man and a machine?

The threat the Turk posed to mankind was eloquently expressed in articles and pamphlets written in an attempt to expose it. The vehemence with which they denied the possibility of machine intelligence points to the panic. “It is quite certain that the operations of the automaton are regulated by mind, and by nothing else,” wrote Poe. “The only question is the manner by which human agency is brought to bear.” “It cannot,” stressed an earlier writer, “usurp and exercise the faculties of the human mind.”

It was crucial, in other words, that a machine should not take over what was rightfully ours. The automaton caused Von Kempelen to be called a “new Prometheus,” a dangerous compliment about playing with fire, and playing with life.

Even, however, after the secret was revealed -- the chest did contain a human operator, hidden in the back half of the box -- people flocked to see the automaton play. More than a piece of mechanics, the Turk was a philosophical toy: neither a man nor a machine but an idea. It was a riddle, and it relied on the public’s desire to be baffled.

It might be argued that that desire is now fulfilled more by humans than by machines. The nearer artificial intelligence experts come to simulating a human being, the more clearly they perceive the particular difficulties of the task; there is still so much we don’t know about ourselves.

In the recent match between Kasparov and Deep Junior, the unknown was represented not by the machine but by the man: How tired will he get, how aggressive will he be, how vulnerable will he be rendered? More than 200 years after the Turk toured the world, human fallibility remains far more gripping than the infinitely mathematical mind of a machine.