A Unifying Bayesian Formulation of Measures of Interpretability in Human-AI
This work addresses the problem of designing coherent interpretable AI systems for researchers and practitioners, though it is incremental as it unifies prior concepts without new empirical results.
The paper tackles the lack of a unified approach for interpretability measures in human-AI interactions by proposing a Bayesian framework that models human beliefs, showing that existing measures like explicability and legibility emerge as special cases.
Existing approaches for generating human-aware agent behaviors have considered different measures of interpretability in isolation. Further, these measures have been studied under differing assumptions, thus precluding the possibility of designing a single framework that captures these measures under the same assumptions. In this paper, we present a unifying Bayesian framework that models a human observer's evolving beliefs about an agent and thereby define the problem of Generalized Human-Aware Planning. We will show that the definitions of interpretability measures like explicability, legibility and predictability from the prior literature fall out as special cases of our general framework. Through this framework, we also bring a previously ignored fact to light that the human-robot interactions are in effect open-world problems, particularly as a result of modeling the human's beliefs over the agent. Since the human may not only hold beliefs unknown to the agent but may also form new hypotheses about the agent when presented with novel or unexpected behaviors.