Maybe, but after all the talk of training and not knowing why decisions are being made, I really want to keep anything resembling emergent at arms length. I very much more interested on multi-agent systems effects on resource allocation and maintainability.

I see your line of thinking as great and interesting, but I'm now in the camp that sees a lot of people blaming "algorithms" for computer decisions that have negative effects (banning or denying resources) on people and thinking that's ok they cannot explain their program's decisions.