Baduk Champ to Face Off Against Proto-Artificial Intelligence

Baduk player Lee Se-dol (left) smiles and hi-fives Google DeepMind CEO Demis Hassabis on the screen during a video conference in Seoul on Monday.

Advertisement

Baduk or go player Lee Se-dol faces off against Google's proto-artificial intelligence program AlphaGo in Seoul in March.

Demis Hassabis of AlphaGo developer Google DeepMind said in a video conference Monday that the match will take place on March 9, 10, 12, 13 and 15 under the Chinese baduk rules.

AlphaGo has been in training for the past 18 months running millions of games with the Chinese rules and could not change the system over a short period, he added.

Lee is prohibited from appearing in any online games until the match with AlphaGo is over. 

If Lee wins all five games against AlphaGo, he will win US$1.25 million. If AlphaGo wins, the money will go to charity.

Lee said, "It is an honor to take part in the first even match between a human pro and a computer. I have been training for one or two hours against AlphaGo every day".

Lee said it took him just five minutes to accept the challenge.

The match will be broadcast live on YouTube, TV Chosun and Baduk TV.

Around 200 journalists came to the video conference, drawn by the hype after AlphaGo defeated the Chinese-born European baduk champion, Fan Hui, in October. Lee is currently the No. 1 player in the world.

Chess computers have been beating people for a long time because there are a limited number of moves they can memorize with sufficient computing power. But baduk is vastly more open-ended and requires the computer to choose among a large number of likely paths to reach the best outcome, a task that some experts believe takes something close to human intelligence.