Islamic Large Language Models: From Knowledge Acquisition to Trustworthy and Hallucination-Resistant AI

The article presents a comprehensive survey on the development of Islamic Large Language Models (LLMs), emphasizing their application in knowledge-intensive domains such as religious and legal question answering. It highlights the necessity for systems that not only understand Arabic but also incorporate curated sources, retrieval-augmented generation, and rigorous evaluation metrics to ensure trustworthiness and reduce hallucinations. This work is significant for practitioners as it outlines the technical requirements and benchmarks needed to create reliable Islamic AI systems that respect the complexities of Islamic jurisprudence and knowledge.