Checkers computer becomes invincible

Daunting task
Arriving at the right solution, Schaeffer said, was akin to accurately mapping out a grid covering Earth’s surface, where every square inch is divided into 1,000 pieces. Each of those pieces, he said, represents a possible checkers position.

Despite the daunting task, Chinook earned some early triumphs for its creators. In 1992, the computer won the right to challenge the reigning checkers world champion, Marion Tinsley, widely considered the best player ever. Chinook lost 4 to 2 with 33 draws in that title match, but won the 1994 rematch by default when Tinsley withdrew due to illness after six draws. It was the first time a computer program had won a human world championship, a feat recognized by the Guinness Book of World Records.

Thereafter, Chinook equaled or bested all of its opponents until its retirement in 1997, the same year that IBM’s chess-playing Deep Blue computer beat reigning world champion Gary Kasparov to far greater fanfare.

Although chess-playing programs have consistently beaten the best players in the world since Deep Blue’s triumph, they cannot claim the invincible status of Chinook. Solving chess, experts say, would require an effort so massive that the world’s fastest computers would need eons to play out every possible move.

Murray Campbell, one of Deep Blue’s designers and programmers, hailed the new checkers solution as “a very important milestone along the way toward machine intelligence.”

Campbell, based at IBM’s Thomas J. Watson Research Center in Hawthorne, N.Y., said the Chinook program uses an accumulated “brute-force intelligence” that could be applied to fields requiring computers to sift through enormous databases. Bioinformatics researchers, for example, often use a brute-force approach when comparing DNA or encoded proteins, while programmers have relied on similar methods to produce more realistic machine-aided translations from, say, English to French.

In 10 years, Chinook and “brute-force intelligence” applications like it may seem trivial, Campbell said. “For now, with what we’ve got available,  it’s quite an impressive accomplishment.”

Next up: poker
Schaeffer isn’t resting on his laurels, however.

Next week, the computer scientist and his colleagues will bring a poker-playing computer program dubbed Polaris to Vancouver, Canada, to challenge two professional players in a $50,000 world championship as part of the annual conference for the Association for the Advancement of Artificial Intelligence.

With games like checkers, Schaeffer said, the information is all out in the open. Not so for poker, when it can be impossible to tell whether your expressionless opponent has an ace of spades or a two of diamonds.

“The real world,” Schaeffer said, “is all about dealing with imperfect information,” meaning that the real test of artificial intelligence’s mettle could arrive in a hard-fought poker game. No one has yet come up with a “superhuman” poker program, though Schaeffer is betting that Polaris may just give some of the world’s best humans a good run for their money.

Don't miss this on MSNBC.com Pass it on: Sons of infertile men may be next The eight biggest Olympic marketing debacles Reality show judges range from wise to wacko Cosmic Log: Is this your jetpack? Forget your inhibitions: World's sexiest islands

MORE FROM FRONTIERS

Tourist or terrorist? Computer knows

Tourist or terrorist? Computer knows Climbing no obstacle for snakebots The next 'American Idol'? Ask your computer How a deadly avalanche can be triggered Green alternative to plastics: liquid wood Getting lost for better architecture Turning air into drinking water Revving up the race for fuel efficiency How to be a disagreeable (but likable) robot Can concept of clean coal be salvaged?   Add Frontiers headlines to your news reader: