From Philosophy to Interfaces: an Explanatory Method and a Tool Inspired by Achinstein's Theory of Explanation

Philosophical theories can guide practical AI explainability: The paper demonstrates that Achinstein's Theory of Explanation from philosophy can be successfully adapted into working software, suggesting that rigorous philosophical frameworks—not just technical approaches—can improve how AI systems communicate with users.

Interactive explanations may be more effective than static ones: The research proposes that explanations work better as interactive processes that answer follow-up questions rather than as one-time static statements, potentially helping users develop deeper understanding of AI decisions.

Preliminary overviews can precede detailed answers: The method treats generic background information as a first step in explanation rather than waiting for users to ask specific questions, which may help non-experts understand AI systems without needing to know what to ask.

The approach was tested on real-world AI systems: The researchers validated their method using IBM's XAI-powered credit approval system, suggesting the framework has practical applicability to consequential AI decisions that affect people's lives (like loan approvals).

Bridging philosophy and interfaces addresses a real gap: The paper identifies and attempts to solve a disconnect between how philosophers think about explanation and how AI systems actually present information to users, which is relevant for making AI more trustworthy and understandable.