Deep Blue beats Kasparov

On 11 May 1997, in a television studio on the 35th floor of a Manhattan office building, Garry Kasparov resigned the sixth game of his rematch with Deep Blue after nineteen moves and lost the match 3.5 to 2.5. He had beaten the machine 4 to 2 the year before. Between the two matches IBM had doubled the number of chess-specific chips, to 480, and hired grandmasters to tune the evaluation. Deep Blue examined about 200 million positions a second and looked twelve or more moves ahead.

It was Shannon's 1950 plan executed with 47 years of Moore's law. There was no learning in it, and almost nothing of the reasoning the AI pioneers had wanted; it was a special-purpose search engine for one game. Kasparov, who suspected human intervention, asked for the logs and a rematch, and IBM dismantled the machine.

The match is the most famous event on this timeline before 2016 because of what people took it to mean: that a machine had out-thought the best human mind at the game that had stood for thinking since the 1950s. Within the field the lesson was narrower, that brute force plus knowledge wins, and Go, where brute force was hopeless, became the new frontier.