<![CDATA[This study examines the implementation of an automated question generation system in the context of Islamic Religious Education (PAI) using LangChain technology and a Gemini-based Large Language Model (LLM). The research methodology includes data collection, needs analysis, system architecture design, implementation, and validation through black-box testing and expert evaluation. The system is designed to automatically generate questions aligned with instructional content and the cognitive levels of Bloom's Taxonomy, ranging from factual knowledge to evaluation and creation. The testing process involved six expert evaluators with educational backgrounds and experience in PAI. Results indicate that the system successfully produced high-quality questions, with an average approval rate of 96.89%. Differentiated scores revealed the highest performance in theoretical interpretation questions (100%) and the lowest in critical analysis (30%), indicating varying system capabilities across the cognitive spectrum. This study demonstrates that integrating artificial intelligence into question generation is highly feasible and yields significant outcomes, although challenges remain in producing questions that stimulate higher-order and authentic thinking.]]>
Copyrights © 2025
