15 years ago today, IBM’s famous chess-playing computer, Deep Blue, beat world champion Garry Kasparov at his own game, making history and changing the way people thought about computers.Murray Campbell was a member of the original team behind the famous chess-playing computer and we had the opportunity to catch up with him this morning.
Campbell is a chess player first and computer scientist second. He became interested in using the game as a way to explore a computer’s capabilities.
As a grad student at Carnegie Mellon University in the 80s, he helped build Deep Thought, a chess computer that was good enough to get IBM’s attention, earning him a job with IBM Research.
He worked with several other engineers and chess players to create Deep Blue, the next-generation chess machine. It made use of parallel processing and custom software to make it a force to be reckoned with. But the first time around, it wasn’t good enough — Garry Kasparov defeated Deep Blue in 1996 by capitalising on its weaknesses as a machine.
Campbell and the rest of the team spent the next year re-tooling the computer, doubling its speed and adding intelligence to help it determine what’s most important in a given game. In short, they taught it how to think more like a human.
“If we had simply used brute processing force, it wouldn’t have had a chance. It needed to be a focused computational effort,” Campbell told us.
Kasparov faced a new and improved Deep Blue the following year. Campbell told us that he was monitoring a metric that showed how well the computer thought it was doing during the game. It climbed steadily and steadily until until Kasparov resigned on the 19th move — Deep Blue had beaten a chess celebrity.
Most recently, IBM Research turned its attention to Jeopardy with WATSON, a computer that processed natural language and searched massive amounts of data to find answers to questions. It handily beat its human opponents.
We asked Campbell where he thinks IBM will set its sights next. He’s unsure, but he suspects “it will involve managing the massive amounts of data generated every day by people, sensors, and cameras, and making sense of it.”