Hot off shutting out the top-ranked player in the world, DeepMind’s Go-playing artificial intelligence AlphaGo is hanging up its hat. The company, which is part of Google parent company Alphabet, announced over the weekend that AlphaGo would be retiring from competitive play.
The announcement comes at the conclusion of a summit in Wuzhen, China that focused on both the game itself and AI. Go is a two-player game of capturing stones that has a large number of possible positions and outcomes — so many that a computer can’t simply crunch through all possible plays like it does in chess. That makes the game perfect for testing algorithms designed to tackle complex problems. AlphaGo faced 19-year-old Chinese Go professional Ke Jie in three matches at the event, and it won all three.
“From the perspectives of human beings, he stretched a little bit and I was surprised,” Ke Jie said at a press conference after the second match. “I also thought I was very close to winning the match in the middle of the game, but that might not have been what AlphaGo was thinking.”
DeepMind isn't totally done with Go, however, as it will release a final academic paper later this year as well as tools that show how AlphaGo analyzes moves.
“The tool will show AlphaGo’s analysis of Go positions, providing an insight into how the program thinks,” the company’s blog post covering all this reads in part, “and hopefully giving all players and fans the opportunity to see the game through the lens of AlphaGo.”
Additionally, DeepMind will publish 50 games showcasing how AlphaGo plays against itself, which is how the AI learns. As for the folks behind AlphaGo, they’ll be moving on to other “grand challenges” — though it’s uncertain exactly what that entails. DeepMind’s post implies that they’ll still be doing work on algorithms and machine learning that could eventually be applied to real-world problems like disease and energy research.
This, of course, has always been the goal. AlphaGo was only ever a means to an end; it just happened to crush its human competition in DeepMind’s pursuit of better and better algorithms. And there’s no better time to retire than on top.
