Former Go champion Lee Sedol still seems to be struggling with AI defeat

Eight years after losing to AlphaGo, former Go world champion Lee Sedol remains deeply affected by AI's capabilities.

Lee saw Go as an art form tied to his character—an extension of his personality that was shattered by the efficiency of an algorithm. Unable to accept defeat, he resigned in 2019, believing AI to be unbeatable and himself to be eternally second-best.

"Losing to A.I., in a sense, meant my entire world was collapsing," said the 41-year-old South Korean in a recent interview with the New York Times.

Today, Lee has mixed feelings about AI's progress. He cautions against underestimating technological advances, drawing from his personal experience. Through lectures, he aims to prepare others for AI's challenges.

"I faced the issues of A.I. early, but it will happen for others," Lee told students and parents at a Seoul education fair. "It may not be a happy ending."

Lee worries that AI could change core human values, which at this point seems like a anecdotal view. "People used to be in awe of creativity, originality and innovation," he says. "But since A.I. came, a lot of that has disappeared."

He advises his 17-year-old daughter to study a field AI can't easily replace. "It’s only a matter of time before A.I. is present everywhere." But he acknowledges AI might create new job opportunities.

Garry Kasparov and Magnus Carlsen, who also stand no chance against AI systems in their sport, take a more measured view. The board game AI AlphaZero, AlphaGo Zero's successor, offers many insightful moves and has transformed chess, they say.

Carlsen even called AlphaZero his hero, and there are more powerful chess systems now. Kasparov said that artificial intelligence hasn't hurt chess, and that people remain interested in human players even when machines are superior.

The reasoning capabilities of systems like AlphaZero have not yet reached LLMs and generative AI - the applications used by the masses. However, Deepmind co-founder Demis Hassabis has hinted that this will happen in future Gemini models. OpenAI is also working to improve the logic capabilities of its models.