Explaining Agent Behavior with Large Language Models
This addresses the need for interpretability in safety-critical agent deployments, such as robots, by providing a method to explain uninterpretable model behavior to humans.
The paper tackles the problem of generating natural language explanations for intelligent agents' decisions using only state-action observations, without needing access to the underlying model. It shows that the approach produces explanations as helpful as those from a human expert, enabling beneficial user interactions like clarification and counterfactual queries.
Intelligent agents such as robots are increasingly deployed in real-world, safety-critical settings. It is vital that these agents are able to explain the reasoning behind their decisions to human counterparts, however, their behavior is often produced by uninterpretable models such as deep neural networks. We propose an approach to generate natural language explanations for an agent's behavior based only on observations of states and actions, agnostic to the underlying model representation. We show how a compact representation of the agent's behavior can be learned and used to produce plausible explanations with minimal hallucination while affording user interaction with a pre-trained large language model. Through user studies and empirical experiments, we show that our approach generates explanations as helpful as those generated by a human domain expert while enabling beneficial interactions such as clarification and counterfactual queries.