How Go Players Disempower Themselves to AI

In March 2016, Google DeepMind’s AlphaGo stunned the world by defeating Lee Sedol, widely considered the strongest Go player, with a 4-1 score. The event "rocked" the Go community, but unlike chess, where computers didn't eliminate human competition, Go initially adapted by integrating AI into commentary and teaching. However, by early 2018, a cheating controversy erupted at the online European Team Championship. Carlo Metta was accused of using Leela 0.11, a predecessor to engines based on AlphaGo's paper, during his games. Accusers claimed his play matched the AI's moves too closely and that his over-the-board performance diverged suspiciously from his online play. The evidence, presented in a sloppy Facebook thread, was criticized as circumstantial. Despite the stigma against AI cheaters, the community backlash led to Metta's exoneration on appeal. However, among non-Italian European players, the belief that Metta used AI in almost every game since 2018 remains widespread, especially given his win rate improvement against opponents of roughly the bot's level.

This incident reveals the deep cultural impact of AI on Go. While top-level human play still thrives, AI tools have introduced new ethical dilemmas. The Metta case shows how the very stigma against cheating can ironically protect alleged cheaters, as the accusation becomes so severe that acquittal seems the only fair outcome. The availability of open-source engines like Leela Zero (released in 2018) has made AI assistance nearly undetectable, forcing governing bodies to rethink rules and enforcement. Go players now navigate a landscape where AI mastery is expected but human authenticity remains valued. The episode underscores broader challenges for competitive games in the age of accessible AI: how to preserve human drama while accepting that computers have surpassed all human abilities. As AI continues to evolve, so must the social contracts that define fair play.