Pro go player beaten by AI

When Deep Blue beat Gary Kasparov in a six-match series in 1997, it was a clear sign that computing power was reaching alarming heights. Now an even more alarming milestone has been reached, as a computer beat a professional Go player for the first time. The result came as a huge surprise at the American Go Congress yesterday, as computers have traditionally struggled to play the ancient Chinese game effectively. The computer program, called MoGo, was being run on a machine that contained 800 processors, operating at 4.7Ghz and peaking at a speed of 15 Teraflop. That equals 15 trillion calculations per second, a speed that helped it run the AI routines to beat 8-dan professional player in a 9-stone handicap game by 1.5 points.

"I can't tell you how amazing this is," said David Doshay, the programmer who suggested the match. He told the American Go E-Journal after the game, "I'm shocked at the result. I really didn't expect the computer to win in a one-hour game." Talking about the impressive progress made, Doshay said, "How much time do we have left? We've improved nine stones in just a year and I suspect the next nine will fall quickly now."

While computers are still a very long way from beating top Go players on equal terms, the win represents a significant leap forward. The huge number of possible moves in Go makes creating an AI routine an enormous task, not to mention the processing power needed to perform such calculations.

Go originated in China some time between 2300 BC and 548 BC, when the first description of a Go match appeared in literature. It was not until 1694 that it was first described in Western literature, and it was not until the 20th century that it became popular in the US and Europe.