E-Prosiding Persidangan Serantau Manasik Haji 2025 (MANASIK2025)

MANASIK2025 Putrajaya, Malaysia, 7-8 Oktober 2025 649 answers in Generative AI strongly affects learning results. This highlights how important good dialogue design and accurate answers are, which is very relevant for guiding people in sensitive rituals. The World Bank’s 2025 study on virtual tutoring with LLMs in Nigeria also shows real learning improvements in secondary schools, A study in Nature Scientific Reports (Bai & Wang, 2025) shows that the quality of interaction and answers in Generative AI strongly affects learning results. This highlights how important good dialogue design and accurate answers are, which is very relevant for guiding people in sensitive rituals. The World Bank’s 2025 study on virtual tutoring with LLMs in Nigeria also shows real learning improvements in secondary schools, proving that these tools can work well in different real-world settings (De Simone et al., 2025). Religious information systems need to focus on being authentic, traceable, and respectful of different schools of thought. A study byMohammed et al. (2025) suggests building a fatwa question – answer system using retrieval-augmented generation (RAG) with large language models (LLMs). This method helps reduce made-up answers and ensures replies follow proper rules, which is especially important for guiding pilgrims with clear references to trusted sources like fiqh manuals and official rulings. At the same time, legal scholars emphasize the need to carefully test AI answers in Islamic law research (Rabb, 2025). They warn against depending too much on AI without expert oversight and stress the importance of citing reliable sources and keeping human review in place. Leaders in this field also call for governance structures, such as an AI Fatwa Council, to make sure values, fairness, and cultural-linguistic sensitivity are protected when AI is used widely. In addition, a national study found that mainstream chatbots like ChatGPT-3.5 generally avoided intolerant answers even in sensitive cases. However, the study also pointed out that domain-specific design and safety guardrails are still needed for religious use (Faizin et al., 2025). There is strong government digital infrastructure for Hajj, growing evidence that GenAI chatbots can improve learning when their answers are accurate and well-designed, and new domain-specific methods such as RAG, provenance, and expert review to handle sensitive religious issues. What is still missing is a complete and tested system that brings these elements together: a multilingual chatbot powered by LLMs and RAG, connected to trusted jurisprudential sources, and built into the Nusuk platform. Such a system could provide both pre-trip learning and real-time guidance during rituals, while ensuring safety, accountability, and inclusivity from the start (Saudi Press Agency [SPA], 2025a; Bai & Wang, 2025; Mohammed et al., 2025). SIGNIFICANCE OF THE STUDY This study carries both practical and scholarly importance. Practically, the proposed Generative AI- integrated chatbot offers a solution to real challenges faced by millions of pilgrims each year. By providing personalized, real-time, and Shariah-compliant guidance, the system has the potential to reduce confusion, prevent mistakes, and ease the emotional burden of performing complex rituals in a crowded environment. The integration of multilingual support also ensures inclusivity, allowing pilgrims from diverse backgrounds to access accurate information in their own language. From a technological perspective, the study contributes to the growing field of domain-specific Generative AI applications. While many AI systems have been developed for education, healthcare, and customer service, limited work has focused on religious practice, especially in contexts as sensitive and large-scale as Hajj and Umrah. By designing a chatbot that combines retrieval- augmented generation, authentic references, and human oversight, this project sets a model for responsible AI use in high-stakes environments. Scholarly significance is also evident in the interdisciplinary nature of the work. It connects Islamic studies, computer science, and human – computer interaction, offering insights into how faith-based