This is incredible: http://poker.srv.ualberta.ca/
Researchers at the University of Alberta have created a computer program that is so good at poker, it may have “solved” the game of fixed limit Texas Hold ‘Em. The most important part is that the program does not win by brute-force calculating all possibilities, but “learns” from trillions of hands, and operates without all the information (unlike Deep Blue, which brute-force calculates the best chess move based on complete game information).
The implications for machine learning and artificial intelligence are fascinating.
Story on the project from the BBC here.