The story about DeepMind learning to be aggressive as it gets smarter is quite interesting and could provide an interesting model of larger interconnected societies. Smaller groups require more civility while larger may not. The question is how to interconnect groups to help cut down on the aggressiveness. Perhaps some of the network ideas in the toy mathematical models from Stuart Kauffman could be useful here? This could be a very interesting problem to work on.
Syndicated copies to: