A System's Approach Taxonomy for User-Centred XAI: A Survey

Explainability research has focused too much on technical methods rather than actual users: The paper identifies that most XAI research takes a "method-centric view" that only makes sense to domain experts, missing the broader goal of helping diverse users understand AI decisions.

There is no standard framework for measuring whether explanations actually work for different people: The survey reveals a critical gap—there are no robust qualitative and quantitative standards to evaluate whether an explanation is suitable for a particular user type, whether they're a developer, regulator, or everyday person affected by an AI system.

A user-centered approach is needed to make XAI truly responsible: The authors propose a new taxonomy based on General System's Theory that evaluates XAI methods based on their appropriateness for different user types, suggesting that responsible AI requires tailoring explanations to who needs to understand them, not just making them technically sophisticated.

One-size-fits-all explanations are insufficient for responsible AI deployment: By emphasizing the need to serve "all user types, including both developers and end users," the paper highlights that explanations must be customized—a technical explanation for an engineer differs fundamentally from what a customer or policymaker needs to understand.