Determining the value function is a difficult problem that is nonetheless key to safely and effectively using reinforcement learning

Brian Christian – The Alignment Problem

The analogies between human and machine learning strategies are skillfully narrated, but rather drawn out.