[justinc-md]

This is exactly the kind of behavior that you'd see with reward hacking in reinforcement learning.

[grp000]

I think humans are pretty good at finding the optimum strategy. It's only really recently that bots have overtaken us in a lot of highly environmentally complex games.