Explainability Auditing for Intelligent Systems: A Rationale for Multi-Disciplinary Perspectives

TL;DR

The paper addresses the need for explainability in AI systems and argues that auditing and certification should incorporate multiple disciplinary perspectives to be effective. It proposes a multi-disciplinary framework—technical, psychological, legal, and ethical—each with distinct dimensions and benefits, to evaluate and communicate system explainability. The authors outline concrete auditing dimensions (e.g., functional, faithful, interactive explainability; understandability, context-dependency, honesty; GDPR, AI Act; responsibility, non-discrimination) and discuss how audits can drive improvement, regulatory compliance, and trustworthy deployment. The work aims to seed practical discussions on developing auditing checklists, normative standards, and potentially normative norms (e.g., DIN) to promote reliable explainability across applied contexts.

Abstract

National and international guidelines for trustworthy artificial intelligence (AI) consider explainability to be a central facet of trustworthy systems. This paper outlines a multi-disciplinary rationale for explainability auditing. Specifically, we propose that explainability auditing can ensure the quality of explainability of systems in applied contexts and can be the basis for certification as a means to communicate whether systems meet certain explainability standards and requirements. Moreover, we emphasize that explainability auditing needs to take a multi-disciplinary perspective, and we provide an overview of four perspectives (technical, psychological, ethical, legal) and their respective benefits with respect to explainability auditing.