“Playing against an AI in hold’em? I think it’s worth a try.”
Lee Sedol, 40, the only human to win a Go match against AI AlphaGo, said this in an interview with the JoongAng Ilbo on Thursday. Lee, who retired as a professional go player in 2019, will pick up poker cards as a hold’em player on Nov. 16. It will be at the “M-Tour” star invitational at the ibex studio in Gwangmyeong, Gyeonggi-do. His opponent is Cha Min-soo, 72, a former professional go player and real-life model of the main character played by actor Lee Byung-hun in the drama “All In.
Lee Sedol said, “I’m still new to the game. I’ve only been playing for a few months, and I recently went to a hold’em pub twice. I’m a 9th dan in Go, but I’m a holin (hold’em + child),” he laughed. “Hold’em and Go are very different, but there are many similarities. I thought there was no other ‘mind sport’ (a sport that uses intelligence to compete) as interesting and fun as Hold’em,” he says. Played with 52 cards, Hold’em is a cerebral game that involves an infinite number of possible bets, as well as bluffing (aka faking it) to trick your opponents into giving up by pretending to have a good hand.
When asked what Hold’em and Go have in common, Lee Sedol said, “They are both abstract strategy games, not simple strategy games. It’s the same thing to play psychological warfare with your opponent and to change your style when your playing style is exposed.” Former Go players Kim Ji-woon and Choi Chul-han are also professional hold’em players. Lee Sedol’s style as a Go knight was the Sili style. This is a style that focuses on building up the house gradually, compared to the power style. “In Hold’em, you have to focus on defense rather than going all-in aggressively at the beginning. You often play defensively and then bet aggressively.”
In 2016, Lee Sedol won his first game against AlphaGo, an AI developed by Google DeepMind, with a score of one hundred and seventy-eight moves after losing three consecutive games. His only loss came against AlphaGo, which was undefeated in 499 games against humans. Looking back, Lee said, “In games 1 and 2, I didn’t know anything about AlphaGo, and in game 3, I broke down psychologically. Game 4 was a lucky win,” he said. “At the time, I could beat AlphaGo because it was a lower version, but now it is virtually unbeatable,” he said. “I learned to play Go with an artistic and academic approach, but when I realized that I could never beat an AI, I began to question the value of that approach. In the end, playing against AlphaGo influenced my decision to retire.”스포츠토토
However, when it comes to AI playing hold’em, “it’s still going to be difficult for AI,” he said. AI needs data to learn, and in the case of Hold’em, there’s not enough publicly available information about players’ play, he explained. However, Isedol says, “If you’re playing against a player who has a lot of experience and has a lot of data, and you’re playing one-on-one instead of many-to-one, then it’s probably worth it. On the other hand, it will be at a disadvantage against players with no data.” His confidence in the AI’s ability to compete against him was also tempered by the fact that “I don’t have a lot of data on me,” he said.