Go Lives (franklantz.substack.com)

Last month, a significant exhibition match took place as top Korean Go player Shin Jinseo faced off against the AI engine KataGo, marking a decade since Lee Sedol's famous showdown with AlphaGo. Shin played with a two-stone handicap, reflecting a minor advantage, yet after losing the initial game, he impressively rebounded to win the next two. This match stands out as it was the first formal contest between a superhuman AI and the world's top human player, highlighting the evolution of AI's capabilities in complex board games. The significance of this event lies in its potential implications for the AI/ML community and the Go world. While many prior AI-versus-human games had been casual, this match was meticulously structured with substantial stakes, including a cash prize and a vehicle for the victor. KataGo's unique strength comes from its innovative design, which allows it to account for and suggest moves based on point estimates, making it not only a formidable opponent but also a valuable educational tool for players. Observers are eager to see how future matches may evolve; with expectations for KataGo to be fine-tuned for greater challenge in subsequent annual encounters, the landscape of AI competition continues to reshape expert play and training methodologies in Go.