“Many have become chess masters; no one has become the master of chess.” So said Siegbert Tarrasch, one of the most eminent players of the 19th and 20th centuries. In the 21st century, however, the foe appears at last to have been overcome, and what’s more it was managed in little more than the time between breakfast and lunch.
Google’s DeepMind artificial intelligence algorithm taught itself hundreds of years of chess knowledge and tactics and then surpassed all previous human invention in the game — all in four hours. Its AlphaZero program was given the basic rules and worked out unique winning strategies by playing games against itself numerous times, according to details published by New York’s Cornell University. The research has yet to be peer reviewed.
Google’s DeepMind division said that it played 100 games against Stockfish 8, the world-leading specialist software, and won or drew them all, with each program given one minute to ponder each move.
Read it all (requires subscription).