Comment by andyxor

Reinforcement learning itself is a dead-end on a road to AI. They seem to slowly starting to realize it, probably ahead of academia.