AlphaGo, Google’s artificially intelligent computer program, beat a professional human Go player last October in 4 out of 5 matches. Go is a very complex board game to master and has been around since the 1000s. The game requires the players to think ahead a few moves and consider what the opponent might or might not do. Tasks that involve a lot of decision-making or snap decisions have typically been difficult for computers to achieve. AlphaGo is helping to break that barrier.
One of the differentiating characteristics of AlphaGo is its ability to sense “good” moves vs “bad” moves. Computers that play chess do so by calculating all of the possible moves of the game, but Go is so complex, the number of options are far too high. With AlphaGo, instead, they loaded a lot of combinations of moves from professional players and allowed it to play thousands of matches against itself to help it learn. Then it played 5 matches against the top Go player in the world, beating him 4 out of 5 times.
You can read more at: