hankyoreh

Go player Lee Se-dol at the press conference after his loss in the first of five games to AlphaGo

After accepting defeat, the grandmaster stayed in his seat and started setting the stones on the board once more to review the decisive moments of the game. Despite his sheepish smile, deep disappointment was evident on his face.

But Lee Se-dol, a 9-dan (the highest level) professional player of the ancient game of Go, had no opponent to review the game with him.

To be sure, there was someone sitting across from during the match on Wednesday: Aja Huang, a 6-dan Go player, who is a British national of Taiwanese descent.

But Aja, an employee of Google DeepMind, was simply placing stones according to the instructions of an artificial intelligence program called AlphaGo. As such, he was in no position to review the game with Lee.

Lee had entered the game room alone, as the champion of mankind against the machines, and after his unexpected defeat, he looked forlorn.

When Lee stepped into the press conference 40 minutes later, he was still obviously experiencing mixed emotions.

“I didn’t think I was going to lose, so this came as a big surprise,” Lee said with a quiet, deliberate laugh.

There had been two surprises for Lee: AlphaGo’s acumen in the opening rounds of play and its mind-blowing moves at decisive movements.

“I had just assumed that AlphaGo would have some trouble at the beginning, but its ability to open up the game was astonishing. And then I thought we were both playing a challenging game of Go when the program blew me away by coming up with these completely impossible moves - that also appeared to be winning moves,” Lee recalled.

In effect, Lee had to face a faceless opponent whose ability wasn’t well known, and he was flabbergasted again and again by the moves this opponent made. The consternation that Lee felt in the match was evident in his remarks.

After watching AlphaGo make the game’s 102th move (counting both white and black stones), Lee mulled over his options for more than 10 minutes.

With black controlling a substantial portion in the center left of the board, AlphaGo had sensed that the game was turning against it and made a decisive move. The move was revolutionary - something that no one had expected, not even Go professionals.

“Even while Lee was playing a close game against AlphaGo, he felt he had an edge, but he seemed stunned by AlphaGo’s strong play on the 102th stone,” said 9-day Go grandmaster Kim Seong-ryong, the official commentator during the match.

“I made some mistakes in my opening moves, and there were also some surprising moves. Still, I think that if I can just compensate for those issues, I‘ll have a chance of winning,” Lee said in regard to his upcoming matches against AlphaGo. “I think my odds of winning now are one to one.”

While Lee could not hide his shock at the day’s defeat, he said that he was relishing the match of a century.

“While it did come as a shock to lose, I really enjoyed the game. I’m also looking forward to the matches to come. I don’t have any regrets about accepting AlphaGo’s challenge. I may have lost today, but I’m confident about tomorrow,” he said.

When asked what AlphaGo means to him, Lee took a moment to think it over, resting his chin on his hand. “AlphaGo provided me with some major surprises, but for now I don’t think I can tell you exactly what it means,” he said.

Lee was unstinting in his praise of Google DeepMind, the company that developed AlphaGo.

Hearing that Google DeepMind CEO Demis Hassabis and AlphaGo programmer David Silver had communicated their respect for him, Lee said he also wanted to express “his deep respect for the two people who made such an amazing program.”

By Kwon Seung-rok, staff reporter

Please direct questions or comments to [english@hani.co.kr]