What's more, go players frequently make moves that to the uninitiated seem like "running away", which is to say not playing near areas of contention. These are placed in moments of low vulnerability so that eventually the player can build towards them. Seldom do even experienced go players know every move that will lead up to connecting with these distant stones, but they know they lend strength to that area and thats a place where they see potential for future moves.
As such, traditional AI game techniques really haven't scaled well to Go. It's only recently that we've began to use huge sums of hardware just to make it play go at a level that most would consider to be an experienced amateur.
It seems like Go is a more recognition-based problem, so we'll have to see if Jeff Hawkin's work might show some progress there once they get their work off the ground.
But maybe it's a good thing we don't have to churn butter anymore, right? Improvements in technology mean that we don't have to deal with the low level stuff that's been done a million times before, allowing us to focus on high level stuff that nobody's done.
"I also suspect Go is related to AGI, so I'm finding the strong advancement a little unnerving (as someone so far skeptical of the likelihood of singularity in our lifetime)"
As I understand it the game is currently playing like a very good amateur player. If the experience with chess is anything to go by, it took an awful lot of work to get from the point of beating most amateurs to the point of beating Kasparov.