London-based artificial intelligence firm DeepMind has developed a computer program called AlphaGo which is capable of beating human players at the ancient Chinese game of Go.
Basic at first glance, Go or weiqi (围棋) is played on a grid with white and black stones which the player places to capture spaces on the board. Though it may seem simple at first, the player has about 200 possible options per move. While playing chess, on the other hand, the player has only about 20 options.
AlphaGo’s first big win came last October, when it beat the European Go champion, Fan Hui. Google’s DeepMind challenged Korean 18-time world title winner Lee Sedol to a showdown against their brainchild. On Friday, AlphaGo defeated Lee for the second time in a best-of-five tourney, a result that shocked the Go world, including the reigning champion, Ke Jie, who claims, “I do not feel the same strong instinct of victory when I play a human player, but I still believe I have the advantage against it. It’s 60 percent in favor of me.”
Like this sharp-shooter, Ke Jie is considered a Go child prodigy, becoming a professional when he was just 11 years old. Now 18, he is the youngest player to win three international championships. He currently holds a win-loss record of 8-2 against Lee. Ke’s competitive attitude has been picqued by the results of this human vs. computer match, expressing his sudden interest to Shanghai Daily:
For me, I’m not interested in playing against an easy opponent, especially for an easy-to-handle computer, for I don’t want to win 100 times out of 100 games, but now I think we can have a try.
The classic game was first referenced back in 548 B.C., possibly as a training method for war strategies. Today, Go is a milestone in the development of artificial intelligence. In 1996, a computer program defeated chess master Garry Kasparov. DeepMind CEO, Demis Hassabis, gives context for Go’s position in the field of game AI development saying,”Since then the really big remaining sort of Holy Grail, if you like, has been Go.”
The only logical opponent left to face off against AlphaGo is Ke. Even he acknowledged the program’s rapid improvement:
According to the pace of AI’s progress, it won’t be long for AlphaGo to beat all human players, it may happen a few years later, even a few months later.
The Economist reports AlphaGo learns by analyzing its opponents’ patterns and sifting through its memory of successful moves in thousands of previous games. This method has apparently proven extremely effective so far. Perhaps lessons learned from AlphaGo can improve Buddhist artificial intelligence.
By Matthew Patel
[Images via CCTV]