AlphaGo Defeats Lee Sedol

AlphaGo defeated Lee Sedol 4–1 in Seoul in March 2016. The result was surprising because many Go experts had expected computers to remain below the strongest human players for years.

The system did not search the full game tree; Go’s branching factor makes that impossible. Instead, neural networks narrowed the search toward plausible moves and estimated the value of positions, while Monte Carlo tree search combined those predictions with simulated exploration. Move 37 in the second game became famous because professional commentators initially considered it strange, yet it proved strategically strong.

Lee’s single victory in game four was also technically interesting: he found a sequence that exposed weaknesses in AlphaGo’s evaluation. The match showed both the power and fragility of learned systems and accelerated interest in deep reinforcement learning far beyond board games.