We observed last year that the game of checkers was solved by a computer. That is, researchers mapped out every possible play in every possible game and determined that perfect play by two players will always result in a draw. A number of games have been solved over the years, but checkers is to date the most complex of these.
Chess has been partially solved, meaning that some variations on the game with a smaller board and / or fewer pieces have been solved, although the full game remains unsolved. There is a big difference, however, between a computer fully solving a game and the same computer being able to beat a human being at that game. For chess, the former is still somewhere in the future, while the latter is a done deal.
There is some debate as to whether machine mastery of games is indicative of any kind of forward progress in artificial intelligence. Heres what the author of the wikipedia general article on chess has to say about the above-linked chess match between Gary Kasparov and Deep Blue:
Garry Kasparov, then ranked number one in the world, lost a match against IBM’s Deep Blue in 1997.[62] Nevertheless, from the point of view of artificial intelligence, chess-playing programs are relatively simple: they essentially explore huge numbers of potential future moves by both players and apply an evaluation function to the resulting positions, an approach described as “brute force” because it relies on the sheer speed of the computer.