Computer Go is actually a huge topic. There are many different solutions, and they compete and still discover new things every year. If you want to get seriousl

Computer Go is actually a huge topic. There are many different solutions, and they compete and still discover new things every year. If you want to get seriously into it you should get in touch with the people who work on it daily. There are people who write doctor theses about parts of a Go AI nowadays and what they create might not even be top notch. When I last checked, it was basically a set of very complex brute force algorithms. What they (need to) do is limit the search space a lot because there are just so many options. And that's where the smartness comes into play, removing subtrees from search space without visiting them. And as far as I know they tried pattern recognition (which is what we humans do) but it didn't work well for AIs. They work more often with density maps for some factors they calculate depending on the position, imho. But all that I know is really only very flat knowledge, and might be wrong.