Aligning Explanations with Human Communication

2025-05-21

AI Summary

Existing interpretability methods neglect user cognitive heterogeneity, yielding explanations poorly tailored to diverse audiences (e.g., clinicians vs. patients). To address this, we propose a listener-adaptive explanation framework that integrates pragmatics-inspired linguistics with the Rational Speech Act (RSA) model—marking the first such synthesis. Our approach models listener cognition without presupposing explicit audience representations, instead learning pairwise explanation preferences from minimal user feedback. Explanations are generated at the conceptual level and optimized for communicative utility. Evaluations across three image classification benchmarks demonstrate significant improvements in explanation–listener alignment. A user study further shows substantial gains in comprehension accuracy and trust among non-expert users. Our core contribution is the establishment of the first pragmatically grounded, listener-adaptive explanation paradigm, enabling semantic alignment between human interpreters and AI-generated explanations.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Computer Vision: Interpretability, Explainability, and Transparency
• Knowledge Representation and Reasoning: Preferences

Application Category

• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Machine learning explainability aims to make the decision-making process of black-box models more transparent by finding the most important input features for a given prediction task. Recent works have proposed composing explanations from semantic concepts (e.g., colors, patterns, shapes) that are inherently interpretable to the user of a model. However, these methods generally ignore the communicative context of explanation-the ability of the user to understand the prediction of the model from the explanation. For example, while a medical doctor might understand an explanation in terms of clinical markers, a patient may need a more accessible explanation to make sense of the same diagnosis. In this paper, we address this gap with listener-adaptive explanations. We propose an iterative procedure grounded in principles of pragmatic reasoning and the rational speech act to generate explanations that maximize communicative utility. Our procedure only needs access to pairwise preferences between candidate explanations, relevant in real-world scenarios where a listener model may not be available. We evaluate our method in image classification tasks, demonstrating improved alignment between explanations and listener preferences across three datasets. Furthermore, we perform a user study that demonstrates our explanations increase communicative utility.

Problem

• Aligning explanations with human communication needs
• Generating listener-adaptive explanations for diverse users
• Improving communicative utility of model explanations

Innovation

• Listener-adaptive explanations using pragmatic reasoning
• Iterative procedure based on pairwise preferences
• Improved alignment with listener preferences demonstrated