Textbook of AI

In May 1997 in New York, IBM's Deep Blue chess computer defeated Garry Kasparov, the reigning world chess champion, in a six-game match by 3.5–2.5. It was the first time a reigning world champion had lost a match to a computer under standard tournament time controls. The result was front-page news worldwide and is generally taken as the moment chess ceased to be a useful benchmark of human-vs-machine reasoning.

Deep Blue was a 32-processor IBM RS/6000 SP with 480 custom chess accelerator chips, evaluating approximately 200 million board positions per second using a parallelised alpha-beta search with extensive opening and endgame books and a hand-tuned evaluation function. Its lineage traced back to Hsu's 1985 ChipTest at Carnegie Mellon, through Deep Thought (1988, the first computer to beat a chess grandmaster in tournament play) to its IBM-funded culmination.

Kasparov contested the result, complaining about restricted access to game logs and suspecting human intervention in one of Deep Blue's moves (now generally understood to have been a bug or a sub-optimal output of Deep Blue's anti-grandmaster opening preparation). IBM dismantled Deep Blue and declined a rematch. The chess-engine race shifted to commodity-hardware programs, by 2005 commercial chess engines on standard PCs surpassed Deep Blue's strength; by 2017 Stockfish was rated several hundred Elo above the human champion.

The Deep Blue match catalysed public discussion of AI capability in a way that earlier symbolic AI achievements had not, and is one of the touchstone moments in the field's public history.