Man and computer live mostly in peaceful coexistence. But in a Philadelphia convention hall this week, it's all-out war. Garry Kasparov, the world's best chess player, is playing the world's best chess computer, ''Deep Blue.''
So intense is the focus on this event - the little-known Association for Computing Machinery, which organized the event, has reaped a publicity bonanza - that it is raising new questions about machine intelligence. If a computer can beat the world chess champion, doesn't that mean it's ''smart?''
It is a fitting time to ask. Exactly 50 years after scientists fired up the first general-purpose computer, one of its descendants has gotten fast enough and good enough to challenge the best chess player in the world. Artificial-intelligence experts have long predicted this moment. Even Mr. Kasparov, normally aggressive and cocksure, uttered some of humanity's concerns.
''Computers play such a huge role in society,'' he said before the tournament. ''But there is a frontier that they must not cross. They must not cross into the area of human creativity.''
Kasparov promptly went on to lose the first game - the first time a machine had ever beaten a reigning world champion in a game played under the rules and time constraints of championship play. Since then, he has fared better, winning Game 2 with the white pieces and playing Deep Blue to a draw in Game 3. Results of the fourth game were not known at press time.
But the questions surrounding machine intelligence won't go away. Although many scientists dismiss Deep Blue as a super-fast calculator, ''that's an awfully human-centric way of viewing intelligence,'' argues Chris-topher Chabris, author of an upcoming book on man-machine relations. ''Maybe it's long past time that people realize that there are other ways of doing things.''
This week's chess tournament is a classic example of man-machine differences. Although artificial-intelligence experts originally believed they could build a machine that ''thought'' like a person, they are now concentrating on building machines with extraordinary calculating capabilities. So, while a chess master might calculate eight moves ahead along one or two lines of play, Deep Blue calculates eight moves ahead for every possible chess move. It decides which move is strongest and can calculate that line of play up to 40 moves ahead with incredible speed.
Meanwhile, Kasparov uses superior chess judgment and a very human ability to grasp the big picture with few calculations. It is this same ability that allows children to know they're seeing, say, a dog, even if they've never seen that particular breed before. A computer would have to process millions of bits to reach the same conclusion.
Although chess experts expect Kasparov to win the match, the days of a human world champion are numbered, they say. ''Sooner or later, you're going to get a computer that 'sees' too much,'' says Dan Edelman, an international chess master and a co-organizer, with Mr. Chabris, of an annual computer-human chess contest known as the Harvard Cup. Already, Deep Blue has improved radically since 1989, when Kasparov handily beat its predecessor.
DEEP BLUE'S current success stems from IBM's ability to get 32 different computers to work together. The idea, called parallel processing, is to break up a large problem that highly specialized processors can work on. By marrying these fast but not very ''smart'' chips with a slower, more flexible supercomputer, the system doesn't waste too much valuable time analyzing useless chess variations. The previous version of Deep Blue had 17 such chips; the current one has 256.
''We've shown this is a good way to do business,'' says Joseph Hoane, advisory programmer at IBM's Watson Research Center in Yorktown Heights, N.Y. The process will probably next be used on a complex business or scientific problem - an example, scientists point out, of how computers and man are cooperating rather than competing.
Nevertheless, this week's chess match also shows how computers are taking over functions and jobs formerly held by people. ''You're getting a little test-tube example here of how a little subculture is reacting to the invasion of technology,'' says Chabris. ''We'll maybe get a little bit better idea of how human-computer relations will evolve.''