Back to News
Release2016-12-21
Faulty reward functions in the wild
Source: OpenAI
Reinforcement learning algorithms can break in surprising, counterintuitive ways. In this post we’ll explore one failure mode, which is where you misspecify your reward function.
openaigpt