Publish date:


Gary Kasparov crushed a computer for all of humanity

Gary Kasparov forever fidgets in his smartly cut tweeds, now drawing his knees up under his chin, now hunching over the chessboard. His opponent, known in chess circles as Deep Thought, betrays no emotion beyond the sickly blue pall that radiates from its squarish face. Deep Thought is not Linda Lovelace's smarter sister but a computer program. And it was being short-circuited at the New York Academy of Art, an art school, in a watershed showdown between man and machine. Or, rather, Soviet world chess champ and machine.

Deep Thought is the reigning computer titleholder, having defeated all comers at the 1989 World Computer Chess Championship in Edmonton. Last fall the program beat a grand master for the first time, bending Bent Larsen of Denmark at a tournament in Long Beach, Calif. A few weeks ago it burned American grand master Robert Byrne. Deep Thought came into Sunday's two-game match with a solid grand master's rating of 2,551, which was well behind Kasparov, whose 2,795 is 15 points above the previous high, set by Bobby Fischer. "The challenge of computer science is to create an artificial intelligence that can beat the world champ," said Hsu Fenghsiung, 30, one of five Carnegie-Mellon graduate students who designed Deep Thought. "It's our Holy Grail."

Kasparov, who is 26, said he was playing the 18-month-old program for the honor of all humanity (plus a $10,000 fee). "I can't visualize living with the knowledge that a computer is stronger than the human mind," he said. "I had to challenge Deep Thought to protect the human race." And if he were to lose? "No one would care about chess anymore. Human chess would become second class."

The first chess-playing machine was a Viennese "automaton," introduced in 1770, that turned out to be a fraud, but electronic gizmos have been playing passable chess since the mid-'60s. Lately, though, they have made a knight's leap in speed and sophistication. Deep Thought is a model of precision. It sifts through the almost infinite possibilities with a thoroughness and speed impossible for the human mind, scanning 720,000 moves a second. "Kasparov only looks at two or three moves," said grand master Edmar Mednis. "He's much more efficient." Deep Thought's big weakness is that it lacks tactical vision and, above all, common sense. "Chess is more than computation," Kasparov said. "You need fantasy and intuition."

Kasparov prepped for Sunday's exhibition by studying 50 of Deep Thought's best games. But the absence of a tangible opponent clearly bothered him. "It's like playing a black hole," he said. "Computers have no energy. Or fear. They fight until the very end and sometimes beyond."

It was a long day for the hundred or so spectators who sat amid the faux Roman statuary flanking the protagonists at the Greenwich Village art school. They had to rely on a large display board for a glimpse of what was going on—when something was going on. Deep Thought's silence made you long for HAL, the computer of 2001: A Space Odyssey, which had a voice like a Wimbledon umpire. HAL had "feelings" programmed into its circuits but lost its head and turned homicidal. Deep Thought has brains but no soul. "It's just a bunch of wires, geometry and silicone," said Hsu.

The match played itself out with a kind of classic inevitability. Deep Thought opened with a pawn to e4. Kasparov countered with the Sicilian defense. Hsu sat by his baby, feeding it Kasparov's moves and watching the ones being considered as they scrolled down the screen. Kasparov eschewed his usual aggressive attack for a slow, positional crush that made Deep Thought's bookish game look positively primitive. Outthought and outmaneuvered, it hung on grimly. "To make Gary keep playing when all is hopeless is almost inhumane," protested organizer Shelby Lyman. Finally, after 2½ hours and 52 moves, Hsu conceded. The second game was even more one-sided. Deep Thought sacrificed a queen early on and fought off one threatened checkmate after another. This time its handlers surrendered after move 37.

"I think the computer needs to be taught one thing," said Kasparov, his domination of chess and the future of humanity secure for now.

What's that?

"How to resign."



Kasparov stifled a yawn as Deep Thought reviewed its options.