Designer-User Communication for XAI: An epistemological approach to discuss XAI design

XAI discussions must include end-users, not just technical experts: The paper identifies a critical gap in current XAI approaches, which predominantly focus on data scientists and ML developers rather than the actual people using AI systems. Meaningful AI explanations require input from all stakeholders early in the design process.

Early involvement of stakeholders improves AI system design: The research emphasizes that discussing XAI scenarios and opportunities should happen during the initial design phase, not as an afterthought. This approach helps ensure AI systems are built with explanations that matter to real users from the start.

Communication between designers and end-users requires structured frameworks: The paper proposes using "Signifying Message" as a conceptual tool to bridge the gap between technical designers and non-technical end-users, suggesting that structured communication methods are necessary to translate complex AI concepts into understandable explanations.

Healthcare AI demonstrates the practical need for better XAI communication: The authors tested their approach using a healthcare AI system as a case study, highlighting that high-stakes domains where AI informs critical decisions (like medical choices) particularly require clear, accessible explanations that end-users can understand and trust.