Who Are You Explaining To? A Multi-Agent System for Audience-Aware XAI Narratives

2026-08-11

AI Summary

Existing explainable AI (XAI) methods, such as SHAP, struggle to deliver explanations that are both faithful to model evidence and appropriately tailored to audiences with diverse professional backgrounds and risk sensitivities—particularly in high-stakes domains like healthcare. To address this gap, this work proposes XstrAI, the first audience-aware, multi-agent XAI narrative framework. Treating local explanations as fixed evidence, XstrAI employs three specialized LLM agents—planning, realization, and verification—that collaboratively generate customized narratives. A bounded revision loop, grounded in inconsistency detection, ensures fidelity and safety. Experiments on diabetes and stroke risk prediction tasks demonstrate that XstrAI significantly outperforms eleven baselines; its explanations are most preferred by both patients and clinicians, remain competitive among data scientists, and enable independent reviewers to accurately identify the intended audience.

Abstract

Feature-attribution methods such as SHAP provide useful evidence about individual model predictions, but their numerical outputs are rarely sufficient for audiences with different expertise, goals, and risks of misinterpretation. In medical AI, the same local explanation must reach patients, clinicians, and data scientists through markedly different forms of communication, and naive verbalization through large language models (LLMs) is prone to weak grounding, conflation of attribution with causal language, and outputs that are persuasive without being faithful to the underlying model evidence. We introduce XstrAI, an audience-aware multi-agent framework that treats local explanations as fixed evidence and structures how it is communicated to each target reader. Each prediction case is encoded as an immutable structured representation, shared identically across audiences so the underlying evidence remains fixed. Generation is factored into three specialized LLM agents responsible for audience-aware planning, linguistic realization, and validation for grounding, attribution consistency, communicative risk, and audience appropriateness, with a bounded revision loop triggered on detected inconsistencies. We evaluate XstrAI on diabetes and stroke risk prediction against 11 baselines, ranging from direct verbalization to a re-implementation of a state-of-the-art narrator. The evaluation combines an intra-narrative regime measuring fidelity to SHAP evidence with an extra-narrative regime assessing audience appropriateness through reference corpora, multi-family LLM judges, and a survey with target readers. In both evaluations, XstrAI's narratives are consistently assigned to their intended audience by independent judges, and preferred over all baselines on Clinician and Patient audiences, with competitive performance on Data Scientist, where audience-conditioned single-prompt baselines lead.

Problem

• Explainable AI
• Audience-aware explanation
• Feature attribution
• Natural language generation
• Multi-agent system

Innovation

• audience-aware XAI
• multi-agent LLM framework
• structured explanation representation
• attribution consistency
• explainable AI narratives