Don't think many have said it here, so just fyi, to me this does not seem like a big 'breakthrough' in AI so much as a good demonstration of smartly combining e

Don't think many have said it here, so just fyi, to me this does not seem like a big 'breakthrough' in AI so much as a good demonstration of smartly combining existing current ideas. Quoting from a developer of a monte-carlo tree based GO bot (http://www.sciencemag.org/news/2016/01/huge-leap-forward-computer-mimics-human-brain-beats-professional-game-go http://www.sciencemag.org/news/2016/01/huge-leap-forward-com...): "Coulom agrees, but he notes that there isn't one key new invention that makes the whole program work. "It's more like a great engineering achievement," he says. "The way they put all the pieces together is really innovative."" Not to disparage the result, but just saying it seems inline with the progress in recent years of using deep neural nets for reinforcement learning with games.