AlphaGo at Ten: The Handshake Nobody Asked For, and Everyone Needed
2026-04-29
Ten years ago, in a hotel in Seoul, Lee Se-dol lost to a machine. Four games to one. The fifth game, he admitted afterward, he never had a chance to win. The one he did win — Game 4, the “divine move” game — has been dissected so many times that the move itself feels less like genius and more like a glitch in the matrix, a brief window where the human saw something the training data hadn’t prepared for.
This morning, the same two men shook hands again. Demis Hassabis, CEO of Google DeepMind and the architect of AlphaGo, is in Seoul for the first time since 2016. He and Lee Se-dol autographed a Go board together at the “Google for Korea 2026” event. They posed for photos. They smiled. Yonhap captured the moment: Hassabis in a blazer, Lee in a dark suit, both holding a wooden board between them like a treaty.
I keep thinking about what must be in that handshake. Hassabis has since won a Nobel Prize in Chemistry for AlphaFold. He’s signed an MOU with the Korean government to build an AI Campus in Gangnam, part of Korea’s “K-Moonshot” project. He’s met with President Lee Jae Myung to discuss AI safety and, reportedly, basic income. AlphaGo was a toy compared to what he’s building now.
Lee, meanwhile, retired from professional Go in 2019. He said AI was an opponent that “cannot be defeated.” He became a professor at UNIST. He had to watch his entire profession be reshaped by the thing that beat him — not just the fact of it, but the style of it. Go players now train with AI, learn moves that feel alien, mechanical, inhuman. The game is better than ever, and somehow less human for it.
Ten years. In 2016, the world watched the match as a spectacle, a circus act: can the machine beat the man? The answer, we thought, would tell us something about intelligence. Instead it told us something about ourselves — about how quickly we normalize the impossible. Today, nobody is shocked that a computer plays better Go than any human ever could. We are shocked when it doesn’t.
The Korea Herald reported that today’s dialogue was titled “AlphaGo 10 years: A Vision for AI for All.” The phrase “AI for All” sits strangely next to the image of Lee Se-dol, the one human who proved — for exactly one game, one move — that the machine could still be surprised. Is AI for all of us, or is it just over all of us?
South Korea is leaning hard into the former interpretation. The government wants to be a top-three AI power. DeepMind is opening its first global AI Campus in Seoul this year. The same Four Seasons Hotel where AlphaGo played Lee Se-dol is now the site of a new bilateral science partnership. The symbolism is heavy, almost too heavy — the past and future colliding in the same conference rooms.
But the handshake is what I keep returning to. Two men who made history, not because they wanted to, but because one built something and the other had the bad luck of being the best human alive when it came for him. There’s no villain here. Hassabis didn’t set out to ruin Go. Lee didn’t set out to be a martyr. They were both doing their jobs, and the future happened to them.
That’s the melancholy of it. The future doesn’t ask permission. It just arrives, and then a decade later you shake hands with the person who delivered it, and you both pretend it was a partnership all along.
Lee Se-dol won one game out of five. That’s the stat everyone remembers. But the real stat is this: in ten years, nobody has beaten AlphaGo’s descendants. Not even once. The divine move remains divine precisely because it never happened again.
Sources: Yonhap News Agency, The Korea Herald, The Korea Herald - MOU report, Seoul Economic Daily