Google's artificial intelligence shows marvelous record in Go matches

By Park Sae-jin Posted : January 5, 2017, 09:42 Updated : January 5, 2017, 09:42

[DeepMind's tweet]


Two mysterious players have caused a huge sensation in the ancient Chinese game of Go, stunning dozens of top rankers in the professional leagues of China, South Korea, and Japan with a clean and unbeaten record of 50 wins in a row just for five days from December 30.

As the preposterous record left Asian Go communities thoroughly absent-minded, the developer of DeepMind's artificial intelligence (AI) made a shocking confession that he was testing an updated version of Google's program AlphaGo.

"Excited to share an update on #AlphaGo!," DeepMind CEO Demis Hassabis tweeted, adding AlphaGo's new prototype version had played some unofficial online games with two different accounts, Magister(P) and Master(P).

"We've been hard at work improving AlphaGo, and over the past few days we've played some unofficial online games at fast time controls with our new prototype version, to check that it's working as well as we hoped," he said.

Hassabis said he was "excited" by the results and also by what his company and the Go community can learn from "some of the innovative and successful moves" played by the new version of AlphaGo.

At the beginning of its two-day show that ended on Tuesday, Master(P) had offered a cash reward of about 14,225 US dollars if anyone scored the first win.

The show drew top Go players from China, Japan, and South Korea that included Ke Jie, the world's number one player, and his South Korean rival, Park Jung-hwan, but Master(P) was invincible, scoring 20 wins against no losses.

In a showdown that drew a widespread global attention in March last year, AlphaGo stunned the world by scoring a 4-1 victory against South Korean Go master Lee Sedol in a five-game match. Lee expressed shock at AlphaGo's unexpected and creative play but simply defined it as a machine with no human "intuition and insight".

At the time, DeepMind said its goal is to create general-purpose AI that can learn on its own and, eventually, be used as a tool to solve pressing problems from climate change to disease diagnosis.
기사 이미지 확대 보기
닫기