Google A.I. Masters the Game of Go

A British company has achieved an artificial intelligence breakthrough as its A.I. software program defeated the European champion of the game Go in a five-game tournament — a feat computer scientists believed to be at least ten years from happening — as reported in the science journal Nature Wednesday.

The program, AlphaGo, beat the Euro champ Fan Hui five games to zero.

The reason why this is recognized as an extraordinary happening is due to the nature of Go. It’s a two-player game, pitting black stones against white stones in a struggle to establish board space and capture opposing pieces by way of encirclement.

Unlike chess or other similar board games, Go is mainly won on instinct, and instinctive play — whereas chess has approximately nine million potential moves, the nonlinear Go offers a possible〖10〗^170 maneuvers — that’s more than there are atoms in the universe.

As the nature of Go is not ruled by logic, experts in computer science had expressed that it would be at least a decade before A.I. would figure out the game so successfully, as Professor Martin Müller of the University of Alberta, Canada told Business Insider, “If you’d asked me last week [about if this could happen] I would have said, ‘No way.’”

AlphaGo, developed by UK tech company Google DeepMind, was trained by having it analyze 30 million moves made by professional human players, playing against itself millions of times, each time learning from mistakes. “It learns what patterns generally occur - what sort are good and what sort are bad. If you like, that’s the part of the program that learns the intuitive part of Go,” said DeepMind’s chief executive Demis Hassabis to the BBC Wednesday, “Now it has all the intuitive knowledge about which positions are good in Go, it can make long-range plans.”

The next challenge for AlphaGo will come in March, when DeepMind is slated to have its software play Go against Lee Sedol, the world’s reigning top Go player.