Brendan Tracey (Santa Fe Institute)
Abstract. Google shocked the go world in January by announcing their program AlphaGo had defeated Fan Hui (2p) in a 5 game series. AlphaGo is currently challenging Lee Sedol (9p) in a 5 game series starting March 9th and concluding March 15th. You may find yourself asking questions like: How has AlphaGo made the strides it has? Why is there a 20-year gap between computers winning at Chess and Go? Why did you put 2p and 9p after the names? Since when is ‘go' more than just a verb?
This will be a very informal seminar describing and contextualizing AlphaGo in the Go community and the AI community. The talk will describe Go as a game, contrast the development of Go AI with the previous efforts for Chess and other games, and discuss some of the advances in AlphaGo algorithm in particular. Finally, I’ll editorialize on why I think this talk obeys Betteridge’s law.
This talk is intended for those who would like more detail than given in the popular press. While it will contain some technical content (and possibly some hastily drawn charts), it’ll be more geared for those who don’t know much about game AI, and will not contain any new SFI research.