AlphaZero
DeepMind's AI for chess, go, shogi. Learned to play from scratch in hours through self-play. Beat the world's best chess engine (Stockfish) in 2017.
Successor to AlphaGo (2016, beat Lee Sedol at go). Difference: AlphaGo was trained on human games, AlphaZero started from zero → discovered new openings and strategies. Algorithm: neural network + Monte Carlo Tree Search. MuZero (2019) took it further — didn't even know the rules but learned them too. Demis Hassabis (DeepMind co-founder) received the Nobel Chemistry 2024 for AlphaFold (protein folding, same architectural base).