Google's AI Retires at the Peak of Go

Google's AI has beaten the world champion three times. It will retire to study other fields.

After two disappointing losses to AlphaGo, world No. 1 Go player Ke Jie entered the final match hoping for an honorable victory. However, the 19-year-old Chinese youth unexpectedly surrendered mid-game, putting an end to the greatest human-AI contest to date.

Ke Jie cried after the match. He used the same strategy as the previous match with good opening moves, creating a good position. But Ke's plan was still ruined by AlphaGo. "It was too perfect," Ke Jie admitted bitterly.

After the match, Ke Jie (center) said that his defeat at the hands of AlphaGo would be the driving force behind his journey to explore the mysteries of Go in the future. Photo: Wired.

At the press conference, Ke Jie apologized for losing, then beat himself up for thinking he could have done better. "I was up against a cold, calm, and intimidating opponent. To the best of my ability, I could only predict half of AlphaGo's moves. I wish I could have done better," Ke Jie said.

Demis Hassabis, founder of DeepMind, the Google subsidiary where AlphaGo was developed, said the artificial intelligence will not play any more games. After defeating the champion, AlphaGo will look for new horizons to conquer.

In addition to Ke Jie, five other ninth-dan players also joined forces to fight AlphaGo but were all defeated. Photo: DeepMind.

Thei Demis, the team behind AlphaGo, will focus on the next challenges to develop advanced algorithms to help scientists solve some of humanity's most complex problems, such as finding new cures for diseases, energy consumption, or inventing revolutionary new materials.

“If AI systems demonstrate that they can discover significant new knowledge and strategies in these areas, it would be truly remarkable. We can’t wait to see what comes next,” Demis said.

After retiring AlphaGo, DeepMind will release data from 50 games AlphaGo played against itself so that the Go community can learn from them. These will be the 50 most intense super games that players have ever seen.

DeepMind will also research and release a teaching tool based on AlphaGo. Nineteen-year-old champion Ke Jie will be working with DeepMind on the project, which Hassabis says will give all players and fans a chance to see the game through AlphaGo's eyes.

DeepMind has no plans to release AlphaGo to the public, but the company is happy to see other companies benefit from its research. Tencent’s Fine Art program and Japan’s DeepZenGo, for example, used similar learning techniques for AI to reach the level of a ninth-dan player.

According to Zing