AlphaGo

What AlphaGo is

Go's search space made brute force impossible, so AlphaGo learned position evaluation and move policies from data and self-play, then searched selectively.

How it works

Policy and value networks guided tree search; later versions learned entirely from self-play with no human games, generalising to chess and shogi.

Why it matters

It demonstrated that learned intuition plus search could exceed human expertise in a domain long considered a benchmark for judgement.

Common uses

• →Teaching reinforcement learning
• →Search plus learning architectures