Year: 2022
Type: preprint
Abstract: Reward hacking -- where RL agents exploit gaps in misspecified reward functions -- has been widely observed, but not yet systematically studied. To understand how reward hacking arises, we construct four RL environments with misspecified rewards. We investigate reward hacking as a function of agent capabilities: model capacity, action space resolution, observation space noise, and training time. More capable agents often exploit reward misspecifications, achieving higher proxy reward and lower true reward than less capable agents. Moreover, we find instances of phase transitions: capability thresholds at which the agent's behavior qualitatively shifts, leading to a sharp decrease in the true reward. Such phase transitions pose challenges to monitoring the safety of ML systems. To address this, we propose an anomaly detection task for aberrant policies and offer several baseline detectors.
Source: arXiv (Cornell University)
Authors: Pan, Alexander, Bhatia, Kush, Steinhardt, Jacob
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4226251505