2016: Gonna Gallantly Go Get Great Go Goal

The game of Go is popular in Western Asia. It’s played on a 19 x 19 board. The rules are simple but give rise to extraordinarily complex play. For many decades, the game received little attention from the AI community because of the enormous search space (much bigger than chess) and the need for extensive Go-specific knowledge. The first Go program was written in 1968, but it wasn’t until 1984 that the computer Go community began holding annual tournaments. This attracted more competitors, and the level of program performance gradually rose. From 1985 to 2000, the Ing Foundation sponsored prizes for programs achieving specified performance levels. When the Ing competitions ended, the programs could still not beat the top 13–15-year-old players despite being given a massive handicap. In the early 2000s, program performance took a big leap forward with the introduction of Monte Carlo Tree Search. In 2015, a project by Google DeepMind used neural networks with reinforcement learning. This, combined with MCTS, produced world-class play. The program, AlphaGo , defeated Go champions Lee Sedol in 2015 and Ke Jie in 2016, two of the greatest players in the history of the game.