<![CDATA[The increasing integration of generative artificial intelligence (GenAI) into educational practice raises critical concerns regarding its capacity to address religious diversity, cultural differences, and ethical complexities in Islamic Religious Education. This research evaluates the multicultural pedagogical quality of AI-generated instructional recommendations and examines the contribution of structured prompting to their contextual and instructional adequacy. Employing a scenario-based comparative design, 288 recorded responses were examined across 24 multicultural dilemmas, six thematic domains, two prompting conditions, and two response profiles. Pedagogical quality was assessed by two recorded raters using six evaluation dimensions, complemented by paired comparisons and categorical analysis of instructional limitations. Structured prompting achieved substantially higher quality scores than standard prompting (77.16 versus 58.80), with a mean difference of 18.36 points (95% CI: 15.82–20.91). Nevertheless, assessment quality remained the weakest dimension, while assessment-related limitations constituted the most frequent deficiency (32.3%). Context-dependent variations further highlighted persistent challenges in addressing religious exclusivism and complex pedagogical dilemmas. These findings suggest that structured pedagogical guidance improves the evaluated completeness of AI-generated recommendations without necessarily ensuring contextual or ethical adequacy. The research underscores the importance of human-guided evaluation, critical pedagogical judgment, and multidimensional accountability in responsible AI-supported multicultural religious education.]]>
