deepmind is not necessary for that and also probably wouldn’t work. If you have player history you can construct perfect models for their playing habits and ev

deepmind is not necessary for that and also probably wouldn’t work. If you have player history you can construct perfect models for their playing habits and even update it as time goes on like a multi armed bandit problem. This can be simple probability heuristics and an optimization function. The problem with it is that the machine becomes biased to past results. Top human players are good because they adapt. I would put my money on the humans