From
Send to

Korean Go player confident of beating Google's AI

Feb. 23, 2016 - 13:03 By KH디지털2

Lee Se-dol, a top-class Go player from South Korea, is known for his exceptional strategy and skills at the ancient board game, but at the same time, he is also known for his provocative comments toward his opponents.

That hasn't changed even though this time, his challenger isn't human.

Lee, who has won 18 world titles, said on Monday that he is confident of beating Google's artificial intelligence (AI) program AlphaGo. In a five-game match set for next month, the 32-year-old South Korean expects to win by a score of 5-0 or 4-1.

"It didn't take long for me to make the decision (to play) because I was so curious about AlphaGo," Lee said Monday. "I only thought about five minutes (before accepting the offer)."

The historic man-versus-computer showdown is scheduled for Mar.

9, 10, 12, 13 and 15 at the Four Seasons Hotel in Seoul. Demis Hassabis, the CEO of the AlphaGo developer Google DeepMind, said the algorithm's programmer will place the markers on the board after watching AlphaGo's moves from a monitor.

The winner of the five-game challenge will receive $1 million. If AlphaGo wins, the money will be donated to UNICEF and other charities.

AlphaGo surprised the Go community after shutting out European Go champion Fan Hui 5-0 last October. But Lee, who went pro at the age of 12, said the result doesn't make him nervous.

"Looking at the match in October, I think (AlphaGo's) level doesn't match mine," Lee said. "Of course, there would have been many updates in the last four or five months, but that isn't enough time to challenge me."

However, Lee, who acquired Go's highest level of ninth dan in 2003, acknowledged that any games against the AI will be very close in a few years.

"If artificial intelligence continues to advance, probably in a year or two years, it will be really hard to guess the results," he said.

Go, known as "baduk" in South Korea, originated in China more than 2,500 years ago. It involves two players alternately putting black and white stones on a checkerboard-like grid. The object is to have larger territories than the opponent by surrounding vacant areas of the board using their stones.

Fellow Go players have predicted that Lee will prevail against AlphaGo.

"I was first surprised that an artificial intelligence program has advanced enough to challenge pro Go players," said Lee Chang-ho, who has won 21 world Go events. "It will be an interesting matchup, but I think Lee Se-dol will win." 

Yoo Chang-hyuk, a ninth dan Go player with 24 titles, including six international tournaments, offered a more careful prediction in the human brain versus computer algorithm matchup.

"Lee Se-dol will have an easy win if AlphaGo's level is the same as what was shown against Fan Hui," Yoo said. "But we don't know how much AlphaGo has progressed since then."

Regardless of the result, scientists are saying the match marks a significant occasion where an AI can challenge a human brain, and it will lead to further advances in the development of human-like algorithms.

"Now, just like humans, an AI can learn from experience and draw inferences to make up strategies," said Jeong Jae-seung, a bio and brain engineering professor at the Korea Advanced Institute of Science and Technology (KAIST). "This matchup will open a new chapter in the history of AI regardless of the result." (Yonhap)