Go, move, shift: Now AI computers are challenging each other

31 Mar 2016

AI has mastered one of the most complex board games on Earth, beating the world’s best at Go. What next? Probably a challenge from another machine.

Just weeks after its overwhelming victory over world-class Go player Lee Sedol, Google’s AI operation AlphaGo is set to be challenged. Not given much time to toast its convincing victory, reports have emerged that the team behind China Computer Go team has Google in its crosshairs, with an AI showdown in the works.

Shanghai Securities News reported that the Chinese team will set a challenge before the year is out, although it’s unclear exactly what that challenge will be. However, it marks a quick deviation away from regular old humans playing Go at the very pinnacle of the game.

After winning 99pc of its games against other Go programmes, last year AlphaGo became the first AI system to beat a pro player when it won out 5-0 against European Go champion Fan Hui.

The bigger challenge, though, was Sedol. That’s because he’s considered to be the best player in the world by some distance. Sedol was confident heading into his March meet-up, which had a $1m prize for the winner of the five-game challenge. He was roundly beaten 4-1, though, saying afterwards that he never wants to play that kind of game again.

So, who else could take on AlphaGo? Well, Facebook’s AI system hasn’t a chance. After all, it was Mark Zuckerberg himself who came out boasting that his company had targeted Go as the ideal place to show off its AI capabilities.

Within 24 hours Google revealed the true extent of its success, though. That, one would have though, leaves little to nobody of note willing to challenge AlphaGo.

There have been suggestions that Google could try its hand at poker, another game of relentless possibilities, however, China’s a big country, with a big interest in Go, so that’s where the next battle will probably be.

Robot image via Shutterstock

Gordon Hunt was a journalist with Silicon Republic