Google AI to play live Go match against world champion

When competing with other available Go programs, AlphaGo won 99pc of its games, and then faced its first real challenge when it competed with European Go champion, Fan Hui, winning 5-0. IBM’s Deep Blue, for example, has beat grandmasters at chess and another IBM creation, Watson, has thoroughly dismantled Jeopardy champions.

The Go-playing program, developed by Google’s London-based artificial intelligence team DeepMind, trounced the best Go-playing AI programs in a series of matches, as reported by Nature last month.

Although it whitewashed its previous professional Go opponent, AlphaGo faces a tougher challenge in the form of Lee Sedol.

The news was announced in a tweet by Demis Hassabis, head of Google’s DeepMind lab. The match will take place between March 9-15, and Google’s Hassabis will reveal more details about the livestream soon.

For those unfamiliar with Go, it is an abstract strategy game played on a grid of black lines, usually 19 by 19. The contest will begin on 9 March and offers a $1m prize.

The London startup, which was acquired by Google for £400 million in 2014, was in a race with Facebook to develop an AI that could defeat a professional Go player.

What is clear is that Google and DeepMind are only getting started: AlphaGo, even if it wins, will be followed by successive AIs that improve on the formula. Past Go robots have only been able to play to an amateur level, but AlphaGo was trained using 30 million moves from games played by human experts.

The difficulty lies in the fact a game like chess has 20 possible moves for the average position. There are roughly 10 to the power of 700 possible board variations for Go, which has made the game hard for computer programmers to crack. DeepMind hopes it can prove that a powerful artificial intelligence is capable of beating the best human player in one of the most mathematically complicated board games ever created.