Google says it has built a system capable of beating the best human players at the ancienct Asian board game Go.
It is a landmark in artificial intelligence (AI), similar to the moment chess champion Garry Kasparov lost to an IBM computer called Deep Blue in 1997.
Go involves placing black and white tiles on a 19x19 board while trying to remove your opponent's pieces.
Because there are so many squares and placing combinations, it's far harder for a computer to master than chess.
On each turn in chess, there are 35 legal moves, whereas in Go there are around 250.
Now Google has revealed that three-time European Go champion Fan Hui was beaten 5-0 in a series of games at the company's headquarters in Kings Cross, London.
The game was played last October, but the full details have only just been released following extensive review.
The software, called AlphaGo, used what Google subsidiary DeepMind calls "deep learning" to build its understanding of the game.
It then picked the moves most likely to lead to victory.
As recently as 2014, Go game AI expert Remi Coulom said it was likely to be another decade before a computer beat the world's top Go players.
0 nhận xét:
Đăng nhận xét