Society

Mar 16, 2016

이세돌 9단과 데미스 허사비스 구글 딥마인드 CEO가 15일 5국이 끝난 뒤, 이번 대국에서 사용된 바둑판을 들고 기념촬영을 하고 있다.

Google DeepMind founder Demis Hassabis (left) and go grandmaster Lee Sedol pose for a picture on March 15. They hold the go board used in the fifth and final game between Lee and the company's computer program AlphaGo.



Five games, one win and four losses.

This is the record of go grandmaster Lee Sedol and is representative of humanity's score against the artificially intelligent computer program AlphaGo.

The five go games notified the world that artificial intelligence is surpassing humans in some regards, and the impact was huge.

The go matches on which the entire world's interest focused have produced a chain of articles that report more than just the game results. They ruminate on the present and the future of AI, and even express vague concerns about a dark future where AI rules over humans, a world often described in science fiction.

However, most observers weren't too negative about the match. The one and only win against the computer program, and the final game after that win, were a stage to prove human creativity, willpower and taste for challenges.

Most go and AI experts before the match foresaw that a dominant human would win. The number of possible moves in a go game is about 10 to the power of 170, so the most logical, mathematic analysis can't always guarantee the best or winning move. This is why it wasn't assumed that the computer program would win against a human opponent armed with his intuition, based on his experience.

The first game was a shocking defeat for the human. The computer program even won the second and third games, showing off its "intelligent" technology, whose level of development was much higher than people assumed. People then bet on AlphaGo winning the fourth and fifth games, too.

이세돌 9단이 15일 ‘세기의 대국’으로 불린 인공지능 알파고와의 다섯 번에 걸친 대국을 모두 마치고 환하게 웃으며 소감을 밝히고 있다.

Go superstar Lee Sedol talks about the match against the computer program AlphaGo and smiles during the press conference after the end of the 'go game of the century' on March 15.



Lee, however, didn't give up. With his strong longing for a win, he finally caused AlphaGo to resign in the fourth game. Lee's win was very much unexpected and all the more valuable, since it was achieved after the program had been getting stronger game after game through self-learning.

Demis Hassabis, the developer of AlphaGo and founder of Google DeepMind, said, "AlphaGo made a mistake." He applauded Lee who only by his fourth game against the computer program was able to exploit AlphaGo's vulnerabilities.

인간과 인공지능의 대결로 전 세계적으로 큰 관심을 받았던 이세돌 9단과 알파고의 모든 대국이 종료된 15일 기자회견장에는 몸을 움직이기 힘들 정도로 많은 기자들이 몰렸다.

Reporters attend the press conference after the five games between Lee Sedol and the AlphaGo computer program. The human vs. AI go matches gained huge attention worldwide.



Lee chose to go for a challenge instead of an easy win for the final game. He requested that he play black in the final game, giving himself a slight disadvantage. The game was tight. It finally finished after 280 moves and the computer was driven into a corner by the countdown clock, unlike in previous games. The go grandmaster lost, but only by one and a half points.

"I feel sorry that the final game ended like this. I wanted it to end well,” said Lee during a post-game interview. "Enjoying the game itself is at the very heart of go, which, to me, hasn't been possible for a very long time. This time, however, I really enjoyed the match against AlphaGo."

By Jeon Han, Chang Iou-chung
Korea.net Staff Writers
Photos: Yonhap News
hanjeon@korea.kr