Go Human-Machine Battle Returns! South Korea's World Champion Shin Jin-seo Loses to AI Model KataGo Despite Two-Handicap Stones

On the 10th anniversary of the historic match between human player Lee Sedol and computer AI AlphaGo, humans and artificial intelligence (AI) once again clashed on the Go board. Shin Jin-seo (9-dan), the world's number one-ranked South Korean Go player, competed against KataGo—the most representative Go AI currently available—in a human-machine showdown hosted by the Korea Baduk Association. In the first match, despite being granted the advantage of two handicap stones, Shin ultimately resigned after 245 moves.

The two-stone handicap was implemented because modern Go AIs, after extensive real-game training and data analysis, have surpassed human players in strength. This format allowed Shin to play as Black and move first, aiming to narrow the gap between human and machine.

Before the match, Shin admitted that it is nearly impossible for humans to defeat top-tier AI at this stage, which is why he chose to challenge under the two-stone handicap. He revealed his strategy was to preserve the initial advantage and avoid complex, aggressive mid-game battles.

However, the game took an unexpected turn from the start. KataGo, playing White, made a rare opening move on its second turn, immediately disrupting Shin's prepared strategy. After the match, Shin acknowledged that while the move might not have been the AI's optimal choice, it was completely outside his preparation, forcing him off his intended rhythm from the very beginning.

During the mid-game, Shin managed to maintain the advantage provided by the handicap for a time. However, a major shift occurred at move 103. When Black attempted to cut through White's formation, KataGo launched an immediate counterattack, successfully breaking into Black's lower-right territory. From that point onward, the game gradually tilted in White's favor. Although Shin tried to mount a comeback, he was unable to close the gap and ultimately resigned at move 245, ending the first game.

South Korean professional Go player Shin Jin-seo. (Screenshot from Hankyung TV)

After the game, during a post-match review, Shin told Korean media that the biggest challenge was the early deviation from his planned opening, which significantly affected his overall strategy. The human-machine match is a best-of-three series, and having lost the first game, Shin will attempt to challenge KataGo again in the next two matches to regain momentum.

Since AlphaGo defeated South Korean professional Lee Sedol in 2016, Go AI models have rapidly evolved. Today, open-source models like KataGo have become essential tools for many professional players in research and training.