Notion of Explainable Artificial Intelligence -- An Empirical Investigation from A Users Perspective

Users prioritize simplicity over technical accuracy: End users want "non-technical and tailor-made explanations" rather than complex algorithmic details, suggesting that explainability standards should be designed for general audiences, not AI experts.

Transparency about data usage is critical: The study found that users require "explanation about personal data usage," indicating that responsible AI must clearly communicate how personal information influences recommendations and algorithmic decisions.

One-size-fits-all explanations don't work: Users want "on-demand supplementary information," meaning people need the ability to customize how much detail they receive based on their individual needs and interests.

Trust depends on reliability and authenticity: Participants emphasized the need for "authentic and reliable explanations," highlighting that users will only trust AI systems if the explanations provided are honest and verifiable, not just plausible-sounding.

Users should shape AI development: The authors propose involving end users directly in the development process for "requirements collection and validation," suggesting that responsible AI requires participatory design rather than top-down implementation of explainability features.