Transforming Islamic Religious Education Assessment through Explainable Artificial Intelligence (XAI): Developing a Transparent and Human-Centered Evaluation Model

Abstract

Artificial Intelligence (AI) has transformed educational assessment by enabling automated evaluation, learning analytics, and predictive decision-making. However, AI implementation in Islamic Religious Education (IRE) assessment remains limited by conventional assessment practices, algorithmic opacity, and inadequate integration of teachers’ professional judgment. This study aims to develop a Human-Centered Explainable Artificial Intelligence (XAI)-Based Assessment Model that promotes transparent, accountable, and evidence-based assessment in IRE. A qualitative library research design with a conceptual model development approach was employed. Data were collected from peer-reviewed journal articles, scholarly books, and international policy documents on AI, Explainable AI, educational assessment, learning analytics, and Islamic educational philosophy. The literature was analyzed through systematic document analysis and interactive qualitative synthesis. The findings reveal that conventional assessment relies on fragmented evidence and subjective judgment, while existing AI models prioritize predictive accuracy over interpretability and accountability. The proposed framework integrates seven components: Learning Data, Machine Learning, Learning Analytics, Explainable AI, Teacher Validation, Professional Judgment, and Continuous Assessment Improvement. The model integrates Islamic values of amanah, 'adl, shūrā, and maslahah, positioning AI as a transparent decision-support system that strengthens teacher-centered assessment rather than replacing professional judgment.