
Shin Jin-seo, right, competes against Go AI KataGo during Game 3 of the Ssen Math-Hankyung Gisin Championship at the Korea Economic Broadcasting studio in central Seoul, Tuesday. Korea Times photo by Yoon Ki-hoon
World No. 1 Go player Shin Jin-seo defeated KataGo, the world's strongest Go artificial intelligence (AI), in a historic exhibition match Tuesday, becoming the first human to win an official match against the AI under the tournament's handicap format.
Shin, 26, defeated KataGo by 11.5 points after 221 moves in the decisive third game of the Ssen Math-Hankyung Gisin Championship, held at the Korea Economic Broadcasting studio in central Seoul.
Although Shin played the three-game series under a two-stone handicap, he rebounded after losing the opening game to claim victories in Games 2 and 3, securing a 2-1 comeback triumph. The result marks the first official series victory by a human player over KataGo under the competition's conditions.
Shin received an appearance fee totaling 150 million won ($108,000) for the three games, along with a 50 million won bonus for each victory, bringing his total prize money to 250 million won. He also received a Genesis G90 luxury sedan for winning two games.
The match was organized to mark the 10th anniversary of the landmark 2016 Google DeepMind Challenge Match between Lee Sedol and AlphaGo, which transformed public perceptions of AI. In that historic match, Lee faced AlphaGo on even terms and lost the five-game series 4-1, with his lone victory becoming one of the defining moments in Go history.
This time, however, the format was reversed: Shin received a two-stone handicap to offset KataGo's overwhelming computational advantage.
Shin entered the deciding game with an estimated 99 percent win probability under the handicap evaluation, equivalent to an 18.5-point advantage. Unlike the first two games, KataGo opened with a different corner strategy, but Shin responded calmly and focused on territorial play rather than aggressive fighting.
The game lasted about three hours and 20 minutes before ending after 221 moves. According to AI win-rate analysis, Shin's chances of victory never dropped below 95 percent throughout the game, making it his most convincing performance of the series.
"I don't think this compares with the one victory that Lee Sedol achieved against AlphaGo 10 years ago," Shin said. "But I believe it still has meaning because it showed, even in a small way, that humans can still hold their own against AI."
Reflecting on KataGo's playing style, Shin said, "AI's weakness seems to be that it is too perfect. Even when it is behind, it doesn't take desperate risks."
He added, "The beauty of human Go is that players can make bold or unexpected moves depending on the situation. I think it was meaningful to show that humans can still compete against AI."
Despite the victory, Shin cautioned against drawing broad conclusions.
"I won the series 2-1, but I still can't say with confidence that I can consistently beat KataGo with a two-stone handicap," he said, adding that he was cautious about discussing whether the handicap should be adjusted in any future rematch.
Before the event, many Go experts had predicted that even a single victory for Shin would be considered a success. Since AlphaGo's emergence in 2016, the gap between elite human players and top Go AIs has continued to widen, with some analysts suggesting that even a six-stone handicap might be necessary.
Shin, however, spent months studying KataGo before the competition and ultimately delivered a result that many in the Go community viewed as a symbolic breakthrough.
Commentator Park Jung-sang, a professional 9-dan player, said the match highlighted how much human players have learned from training with AI over the past decade.
"If humans hadn't continued learning over the past 10 years, even a three-stone handicap would have been difficult," Park said. "By constantly training with AI, which has become part of everyday life, humans have continued to evolve. This match proved just how remarkable human strategy can be."
This article from the Hankook Ilbo, the sister publication of The Korea Times, is translated by a generative AI system and edited by The Korea Times.