Clash of the Explainers: Argumentation for Context-Appropriate Explanations

One-size-fits-all explanations don't work: The paper demonstrates that no single XAI technique is universally best, meaning organizations need to match explanation methods to specific contexts rather than applying the same approach to all stakeholders.

Stakeholder characteristics should drive explanation choice: Explanations must be tailored to who is receiving them—different stakeholders (e.g., regulators, end-users, domain experts) have different needs and backgrounds that should determine which explanation technique is used.

Argumentation can make explainer selection transparent: By using formal argumentation techniques to reason through which explainer is most appropriate, the system can show why a particular explanation method was chosen, adding a layer of transparency to the AI explanation process itself.

A modular system maps stakeholder needs to explanation methods: The proposed framework connects stakeholder characteristics (their knowledge, role, and needs) to the strengths and weaknesses of different explanation techniques, enabling systematic rather than arbitrary selection decisions.

Transparency in the selection process builds trust: By formalizing the reasoning behind which explanation technique gets used for which stakeholder, organizations can demonstrate that explanation choices are deliberate and justified, rather than appearing arbitrary or self-serving.