This week a computer defeated a professional player human player at the game of Go. As I mentioned in my previous article, The Second Half of the Chessboard, computers have already defeated the best humans at chess and jeopardy. Winning it Go is next milestone in artificial intelligence.
The team from Google’s Deepmind has developed a program, called AlphaGo, capable of beating a human player at the game of Go. AlphaGo uses a type of artificial intelligence called deep learning, which uses neural networks to play millions of games and then learn the best strategies. They challenged Fan Hui, the current European Champion.
In interviews Fan Hui said that if he had not been told that he was playing a computer, he would have thought that he was playing a real person (implying that AlphaGo would have passed this low level turing test). Fan Hui also stated that after the first game he thought that AlphaGo would play a wide, spread out strategy so he changed his strategy to fight more, as you can see below. Fan Hui lost all five games.
Alpha build challenge Lee Sedol in March.
Below is a record of each game of Fan Hui vs AlphaGo.