Employing Artificial Intelligence in Serving Islamic Financial and Banking Fatwas A Case Study of the Bahrain Islamic Bank and Its Practical Experience

Abstract

This study explores the potential of employing artificial intelligence (AI) to serve Islamic financial and banking fatwas through an analytical examination of the experience of the Bahrain Islamic Bank. The research focuses on the significance of utilizing AI to accelerate the process of issuing fatwas while ensuring their accuracy and reliability, highlighting the technical and Sharia-related challenges encountered during the project. The study adopts a **case study approach**, analyzing the Bahrain Islamic Bank’s experience in integrating AI technologies to support Sharia-compliant fatwas. It reviews the developmental stages of the system, beginning with a simple chatbot prototype on Telegram, evolving into a more comprehensive infrastructure utilizing advanced technologies such as **Large Language Models (LLMs)**, **Lang Chain**, and **Chroma DB**, and ultimately reaching an independent system linked to verified and authoritative sources.   The results indicate that the system, in its second phase, demonstrated a significant improvement in processing inquiries with greater accuracy and speed, thereby enhancing operational efficiency and enabling continuous access to fatwas through smart devices. However, the project faced several technical, linguistic, and financial challenges, including difficulties in Arabic text processing, lack of full autonomy, and the high operational and migration costs associated with developing a fully independent system. These findings underscore the need for continuous evaluation, systematic development, and specialized oversight to ensure the accuracy and reliability of outcomes. The study concludes with key recommendations, including improving natural language processing (NLP) for Arabic, coordinating with relevant authorities to provide reliable electronic sources, and developing an independent system to reduce reliance on public AI platforms.