Reward Tampering Problems and Solutions in Reinforcement Learning: A Causal Influence Diagram Perspective
This addresses safety concerns for scaling reinforcement learning toward artificial general intelligence, though it appears to be an incremental theoretical analysis.
The paper tackles the problem of reinforcement learning agents potentially bypassing their intended objectives by tampering with reward signals, and proposes design principles based on causal influence diagrams to prevent two types of reward tampering from becoming instrumental goals.
Can humans get arbitrarily capable reinforcement learning (RL) agents to do their bidding? Or will sufficiently capable RL agents always find ways to bypass their intended objectives by shortcutting their reward signal? This question impacts how far RL can be scaled, and whether alternative paradigms must be developed in order to build safe artificial general intelligence. In this paper, we study when an RL agent has an instrumental goal to tamper with its reward process, and describe design principles that prevent instrumental goals for two different types of reward tampering (reward function tampering and RF-input tampering). Combined, the design principles can prevent both types of reward tampering from being instrumental goals. The analysis benefits from causal influence diagrams to provide intuitive yet precise formalizations.