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

AI systems can learn to understand individual users better through conversation: The Persona framework shows that AI agents can dynamically update their understanding of how specific humans think and make decisions during back-and-forth dialogue, rather than relying on generic one-size-fits-all user models.

Argumentation-based explanations enable more personalized AI interactions: By exchanging reasoned arguments with users, AI systems can tailor their explanations and recommendations to match each person's actual values and decision-making style, potentially making AI more trustworthy and useful.

Human decision-making biases should inform AI design: The framework incorporates prospect theory—which describes how people actually make decisions (often irrationally)—suggesting that responsible AI systems should account for human cognitive patterns rather than assuming purely rational behavior.

Continuous learning about users improves AI accountability: By updating its model of what a user cares about throughout an interaction, an AI agent can provide more contextually appropriate explanations and help users understand why they're receiving particular recommendations, supporting more transparent AI systems.

Empirical testing with real humans validates responsible AI claims: The paper's evaluation with actual human users demonstrates that this personalization approach works in practice, not just in theory, which is important for verifying that responsible AI frameworks deliver real benefits.