Personalizing explanations of AI-driven hints to users' characteristics: an empirical evaluation

Personalized explanations can improve AI transparency for specific user groups: The study found that tailoring how AI systems explain their decisions—rather than using one-size-fits-all explanations—increased engagement with explanations among students who typically avoid them, demonstrating that explanation effectiveness depends on matching the presentation to individual characteristics.

AI explanations matter most for users who naturally disengage: Students with lower "Need for Cognition" (preference for thinking deeply) and lower "Conscientiousness" (attention to detail) were less likely to seek explanations on their own, but when explanations were personalized to their traits, they interacted with them more and learned better—suggesting personalization can help reach users who would otherwise ignore AI justifications.

Responsible AI requires understanding diverse user needs: The research highlights that a single explanation format may inadvertently exclude certain users from understanding how AI systems work. By recognizing that different people have different cognitive preferences, organizations can design more inclusive AI systems that serve all users, not just those naturally inclined to seek detailed information.

Personalized explainable AI shows measurable educational benefits: The formal user study demonstrated that personalization improved three key outcomes simultaneously: students' interaction with explanations, their comprehension of the hints, and their actual learning performance—providing concrete evidence that PXAI isn't just theoretically sound but practically valuable.