<![CDATA[Background: The rapid adoption of generative AI tools for self-directed learning raises concerns about content accuracy in specialized knowledge domains. In Islamic studies, where epistemic authenticity depends on authenticated transmission chains (sanad) and established jurisprudential methodologies, the validity of AI-generated religious content remains largely unexamined. Objective: This study evaluates the validity and information adequacy of ChatGPT-3.5 responses on waqf (Islamic endowment) for early learners without formal Islamic education. Methodology: A qualitative content analysis was conducted using 14 Indonesian-language prompts across four waqf dimensions: basic concepts, pillars and conditions, nazir (trustee), and contemporary issues. Responses were compared against three authoritative references using a five-point rubric-based scoring system. Results: ChatGPT achieved an average validity score of 4.48 and adequacy of 4.03 out of five. Performance peaked in basic concepts (validity 5.00; adequacy 4.67) but declined sharply in contemporary issues (3.50; 2.50), where oversimplifications, inaccurate madhhab attributions, and procedural errors emerged. Conclusion: ChatGPT provides accurate foundational waqf knowledge but lacks the juridical depth and contextual precision required for intermediate or advanced Islamic legal study. Implications: AI tools should function as supplementary learning aids rather than substitutes for sanad-based Islamic education. Digital religious literacy programs and shariah-informed AI development are needed to mitigate epistemic risks in religious knowledge acquisition.]]>
Copyrights © 2025
