Hands-On Artificial Intelligence for Beginners

AlphaGo and intelligent game-playing AIs

While MCTS has been a cornerstone of game-playing AI for a while, it was DeepMind's AlphaGo program that really took game-playing AIs into the modern age. AlphaGo and its derivatives (AlphaGo Zero and AlphaGo Master) are game-playing AI systems that utilize MCTS to play the notoriously difficult ancient Chinese game of Go. With 10⁷⁶¹ possible game combinations, creating a system to play the game became something of a milestone in the AI world. It's even the subject of a much talked about documentary by the same name.

AlphaGo uses a combination of MCTS with deep learning methods that made the Alpha Go programs truly extraordinary. DeepMind trained deep neural networks, such as the ones that we have ...