E-Prosiding Persidangan Serantau Manasik Haji 2025 (MANASIK2025)

MANASIK2025 Putrajaya, Malaysia, 7-8 Oktober 2025 648 BACKGROUND The preparation for Hajj and Umrah has traditionally relied on physical training courses, printed manuals, and face-to-face instruction by religious teachers. While these methods provide foundational knowledge, they are limited in reach and often fail to address the diverse needs of global pilgrims. With increasing numbers of pilgrims coming from non-Arabic speaking countries, language barriers frequently complicate learning. Moreover, many pilgrims attend training sessions months before their journey and may struggle to recall details during the actual rituals. Over the past decade, digital technologies have been increasingly adopted to address these challenges. Mobile applications have been developed to provide step-by- step guides, du‘a collections, maps, and translations. Examples include official platforms such as Nusuk and community-developed apps focusing on Hajj logistics and supplications. However, most of these tools remain static and reference-based, offering information in a one-way format rather than interactive, context-sensitive guidance. Studies show that while these apps are helpful, they lack adaptability, real-time responsiveness, and personalization. The lived experience of pilgrims highlights the seriousness of these limitations. Surveys have revealed that many pilgrims’ experience confusion during key rituals, such as Tawaf or Sa‘i, particularly in overcrowded conditions. Testimonies also describe situations where pilgrims were unable to obtain accurate religious advice when they needed it most, either because their guides were unavailable or not sufficiently qualified. In some cases, incorrect advice led to mistakes in performing rites, resulting in anxiety, doubts, and the need to repeat certain rituals. These challenges underscore the urgent need for a smarter, more reliable support system. Generative AI technologies, particularly Large Language Models (LLMs), offer strong potential to address this gap. When combined with an authentic knowledge base derived from the Qur’an, Hadith, and fiqh sources, LLM-powered chatbots can deliver interactive, real-time, and multilingual guidance. Unlike static applications, they are capable of understanding diverse questions, adapting to context, and providing tailored support during critical moments. LITERATURE REVIEW The Arab government reports and news in 2025 show that pilgrimage services are becoming quickly digital. The Nusuk platform has been built as a one-stop hub, and new AI assistants and service robots are being used to answer questions in many languages, give on-site information, and even help with recitation. These steps show that there is already a system where smart tutoring and support agents could be added (Saudi Press Agency [SPA], 2025a; Arab News, 2025). Recent news about “fatwa robots” and technologies for managing crowd flow highlights this progress bu t also brings up questions about religious authority and governance (Daily Sabah, 2025). At the same time, health research on Hajj in the year 2025 points out how AI can support mass gatherings through triage, guidance, and decision-making. This role is very similar to how AI could also guide pilgrims in performing their rites (Alquayt et al., 2025). Recent reviews and reports agree that large language models (LLMs) can support learning by giving feedback, personalizing help, and guiding students in many fields. Frontiers in Computer Science explains different ways to evaluate LLMs such as their abilities, setups, and parameters; and stresses that assessments should match the learning task to ensure responsible use (Saleh et al., 2025). A systematic review in Springer focused on education shows that LLMs can help with teaching tasks like drafting, giving feedback, and formative assessment, and that they work best when linked closely to the curriculum (Guizani et al., 2025). Reports from the SIGCSE 2025 conference also highlight benefits in coding and computer science education but warn that integrity and teaching quality must come first (Raihan et al., 2025). A study in Nature Scientific Reports (Bai & Wang, 2025) shows that the quality of interaction and