Artificial intelligence · 2006
Monte Carlo tree search
Random playouts guide search, and computer Go suddenly improves.
Remi Coulom named Monte Carlo tree search in 2006, and Kocsis and Szepesvari's UCT gave it a principled exploration rule. MCTS lifted Go programs to strong amateur level and later formed the search half of AlphaGo.
What it built on 2
What it led to 1
Related measurements
The dashed line marks when it first worked.
Training compute of notable AI models Source: Epoch AI, Notable AI Models (CC BY 4.0)
ML hardware FLOP/s per dollar Source: Epoch AI, ML hardware (CC BY 4.0)