Deep reinforcement learning from human preferences

Christiano, Paul, Leike, Jan, Brown, Tom B. · arXiv · 2017