I wonder if you could apply reinforcement learning to convincing humans of things.

Perhaps something best not attempted...