Dear Readers,

Ten years ago, a machine placed a single stone on a Go board, and the world hasn't been the same since. In this week's DeepDive, we trace how AlphaGo's legendary victory over Lee Sedol in March 2016 sparked a chain reaction that led to a Nobel Prize, shattered mathematical records untouched for half a century, and laid the foundation for what Google DeepMind now openly calls its path to AGI. It's a story about creativity, self-learning systems, and the surprisingly thin line between a board game and the frontiers of science. Plus, I share my take on why reinforcement learning, not larger language models, might be the real key to building artificial general intelligence.

All the best,

On March 9, 2016, more than 200 million people around the world tuned in to watch something most of them had never cared about before: a board game. In a sleek conference room at the Four Seasons Hotel in Seoul, a quiet 33-year-old Go champion named Lee Sedol sat across from an empty chair. His opponent wasn’t in the room - it was running on a bank of servers thousands of miles away. The AI system was called AlphaGo, built by a then-small London startup named DeepMind. And by the time the match was over, the world of artificial intelligence had been turned on its head.

Go is not chess. It is far more complex, with more possible board positions than there are atoms in the observable universe. For decades, AI researchers considered it the ultimate frontier — a game so rooted in human intuition and pattern recognition that no machine could reasonably master it. Most experts believed we were at least ten to twenty years away from that milestone. AlphaGo shattered that timeline in a single week.

But the real legacy of AlphaGo isn’t about Go at all. It’s about what happened after the last stone was placed. In the ten years since that match, the techniques pioneered by AlphaGo have produced a Nobel Prize in Chemistry, cracked mathematical problems unsolved for over half a century, and set the stage for what may become the most consequential technology in human history. The question is no longer whether a machine can beat a human at a board game. The question is: can the same creative spark that produced AlphaGo’s legendary Move 37 ultimately help us understand the fundamental mysteries of nature - and perhaps even build a truly general intelligence?