Google’s AI program AlphaGo whipped the world’s top human Go player, Ke Jie, so resoundingly this spring that he said he wouldn’t ever play a machine again.
Now, he’s coming back for more.
Except this time, Jie won’t be facing any smart machine from Google’s AI research lab, DeepMind. Instead, Reuters reports he’ll take on AI opponents from China, Japan, and Taiwan. The match will take place in April 2018.
One of Jie’s opponents will be an AI program from Tencent, a Chinese tech powerhouse that has been deeply invested in its own AI development. Jie will also square off against Japan’s DeepZenGo and Taiwan’s CGI.
Go is an ancient, sophisticated Chinese board game that has been played for well over 2,000 years. The number of legal moves has been compared to “the number of atoms in the universe,” so any machine — whether human or artificial — has a maddeningly complex strategy to consider. The goal in the two-person game is to surround — and conquer — your opponent’s tiles.
The reason why the nineteen-year-old Jie isn’t challenging DeepMind to a rematch is likely because AlphaGo is simply not available. Google’s AI lab retired their advanced algorithm-crunching machine after it whipped Jie this past spring, 3 – 0.
Following AlphaGo’s public retirement, however, the DeepMind lab pitted the program against an even more advanced machine, AlphaGo Zero. AlphaGo Zero’s victory turned out to be one of AI’s most impressive accomplishments of 2017. DeepMind gave AlphaGo Zero the ability to teach itself how to beat formidable AI opponents, with no previous ideas about game strategy. The program, then, taught itself to became dominant. In 40 days, AlphaGo Zero had beaten every previous version of AlphaGo — including the version that toppled Jie.