I have played Go and I understand that it is a complex and nuanced game. However, I also understand that it's a game very heavily based on patterns and pattern-matching.

Have you read the research paper and seen how much of AlphaGo's architecture was designed around the specifics of Go gameplay? (Including a set of handcrafted Go-specific training features.)