The folks at Deep Mind are at it again. Recently they turned their attention to chess and came up with an AI which very quickly beat up on the previously best chess computers. Now they’ve come up with a Go teaching tool, focusing on the opening.
Former pro and YouTube content creator Haijin Lee has posted an excellent brief video on her channel introducing the tool.
Deep Mind has provided win-rates for various moves in common openings based on AlphaGo evaluation. It’s a fascinating tool, but it isn’t clear to me yet how valuable it is going to be for weakling amateurs like me. After all, I’m capable of screwing up a promising game with just a few remarkably bad moves anywhere through the middle game, and I can lose considerable ground in the endgame to a stronger player.
I’m not saying having a strong opening is not important. Of course it is, but there are so many aspects to the game – building, reducing, invading, fighting, life & death, direction of play, maintaining sente and so on. Still, improving opening play has to be a good thing and at some point along the way I’m going to play around with the tool and see if I can improve in that area.
Just out of curiosity – are there any go players reading this blog? Have you been following the work Deep Mind has been doing with AI development?