AlphaGo is a combination of symbolic AI (similar to Deepblue) to search for best hands, and neural network to recognize a localized winning pattern. The introduction of deeplearning is to compensate the inability of the searching from "unlimited" possibilities.