Does Your AI Agent Get You? A Personalizable Framework for Approximating Human Models from Argumentation-based Dialogue Traces

2025-02-22

AI Summary

Existing explainable AI (XAI) methods predominantly rely on static, pre-specified human user models, limiting their ability to dynamically adapt to evolving user cognition and preferences during interaction. To address this, we propose Persona, a novel framework introducing the first online user modeling mechanism that jointly integrates prospect theory and Bayesian belief updating. Persona infers and incrementally refines a distribution over user models in real time through argumentative dialogue trajectories. It formalizes cognitive biases via probability weighting functions, encodes logical reasoning using argument graph structures, and enables continual learning from natural-language debates. In empirical studies with real users, Persona achieves statistically significant improvements in belief modeling accuracy (p < 0.01) and generates more natural, personalized explanation interactions. It outperforms state-of-the-art XAI methods across multiple quantitative and qualitative evaluation metrics.

Technology Category

• Humans and AI: Explainable AI (XAI) for Human Understanding
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Knowledge Representation and Reasoning: Reasoning with Beliefs

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

Explainable AI is increasingly employing argumentation methods to facilitate interactive explanations between AI agents and human users. While existing approaches typically rely on predetermined human user models, there remains a critical gap in dynamically learning and updating these models during interactions. In this paper, we present a framework that enables AI agents to adapt their understanding of human users through argumentation-based dialogues. Our approach, called Persona, draws on prospect theory and integrates a probability weighting function with a Bayesian belief update mechanism that refines a probability distribution over possible human models based on exchanged arguments. Through empirical evaluations with human users in an applied argumentation setting, we demonstrate that Persona effectively captures evolving human beliefs, facilitates personalized interactions, and outperforms state-of-the-art methods.

Problem

• Dynamic learning of human models
• Personalized AI interactions
• Argumentation-based dialogue adaptation

Innovation

• Dynamic human model learning
• Argumentation-based dialogue adaptation
• Bayesian belief update mechanism