"It Felt Like Playing Tennis Against a Wall"... Lee Sedol Reflects on His Match Against AlphaGo

"Need AI Technology... The Key Is Speed Control"
"Learning Baduk from AI Feels Like Looking at the Answer Sheet"

Lee Sedol 9-dan, who had a historic match against the AI AlphaGo in 2016, shared his views on the development of AI.

In an interview with Google Korea on the 19th, Lee Sedol recalled his match with AlphaGo, saying, "In Go, we often use the term 'game rhythm,' but (AlphaGo) had none of that," and added, "It felt like hitting tennis against a wall. Also, it played so well that it created a sense of alienation."

Previously, Lee Sedol suffered three consecutive defeats against Google DeepMind's Go program AlphaGo in 2016 but secured a victory in the fourth game. Lee Sedol's single win remains the only defeat AlphaGo has ever experienced.

9-dan Lee Sedol. [Image source=Google Korea YouTube]

Lee Sedol said, "After losing three games, especially feeling mentally exhausted after the third game, I actually felt more at ease," and added, "I think I saved face by winning the fourth game." He continued, "At that time, there were many supportive messages like 'Hang in there,' and 'The remaining games will be tough, but do your best.' Those messages gave me great strength."

Lee Sedol mentioned that the emergence of AI has changed the way people learn Go. He said, "The Go I learned before was an art where you ponder alone and then two people meet to create a single masterpiece. But at some point, watching AI, learning Go felt like looking at an answer sheet. Of course, there are good aspects, but there are many parts I feel regretful about."

Regarding AI technology, Lee Sedol said, "The most important thing is speed control," and added, "We need AI technology to live our lives. However, it must be developed with well-controlled speed and under clear principles and standards."

When asked if he would learn Go again if he were born again, he said, "I could learn Go. I could enjoy it, but if you ask whether I would become a professional Go player, my thoughts are different." He continued, "In the past, I would have said positively that I would learn Go again and definitely become a professional player, but after AlphaGo appeared, my thoughts changed. I would learn Go, but perhaps the other side, making AI, would also be fine," he said with a smile.