rmoehn comments on Model Mis-specification and Inverse Reinforcement Learning

[rmoehn]

Humans can be assigned any values whatsoever… is a great basis for understanding the last section of this article.