It is not clear at all that the cooperating team is more difficult to beatä
> * solving Go (hard) Not only hard, but 'solving Go' it is beyond AI, and probably cannot be done.
* finding ways to translate insights gained from observing strong Go AI into real-life situations relevant outside of Go (my personal favourite; Go has that fascinating simple-meets-complex abstracting property [1])
Here I fully agree, as conan noted in another thread, stylistic factors are quite important, and the art-like approach is a thing in Go, so I would love to know "what the AI thinks about it" :)
1. Perfect play from any game position.
2. Proof of perfect play.
#1 may be achievable with a sufficiently clever algorithm. (We may suspect it is perfect and may only later prove it.) #2 may be much much harder. Neither are suitable for brute force approaches to play or proof as some problem spaces are.
I mean, it's mildly interesting that DeepZen uses fewer resources (just ~40 CPUs and 4 GPU, compared to AlphaGo's ~2k CPU and ~200 GPUs, IIRC). But that's relevant only to hardcore Go enthusiasts who would want to run such AIs on their own hardware.
In contrast, the next interesting milestones in Go AI are:
* AI beating a team of cooperating top pros, say top-10 humans vs AI (easy)
* solving Go (hard)
* finding ways to translate insights gained from observing strong Go AI into real-life situations relevant outside of Go (my personal favourite; Go has that fascinating simple-meets-complex abstracting property [1])
[1] https://rare-technologies.com/go_games_life/