Degree

Master of Science in Islamic Banking & Finance

Department

Department of Finance

Faculty/ School

School of Business Studies (SBS)

Date of Submission

Spring 2024

Supervisor

Dr. Muhammad Asif Jaffer, Assistant Professor, Department of Accounting & Law

Keywords

Generative AI, Large Language Model, Islamic Banking, Shariah Advisory

Abstract

Abstract Background It is mandatory for Islamic banks to adhere strictly with Shariah principles in all their operations, and offerings raising the demand for experts which have proficiency in banking and financing related Shariah knowledge. In this regards, the capability of Large Language Models (LLMs) have been explored in complex financing scenarios. Objective This exploratory study evaluates the capacity of LLMs, specifically ChatGPT, Gemini, and Meta AI in Shariah advisory for complex financing scenarios within Islamic banking. The capability of these models were investigated against three benchmarks; identification of Shariah compliance issues, providing Shariah basis/references, and offer Shariah guidance on handling the issue. Methodology Qualitative approach is used in this study by complex realistic financing scenario creation, expert consultation, and LLM evaluation. Ten different Islamic banking financing scenarios were developed, and LLMs were tasked with analyzing these scenarios across three benchmarks listed above. Then the LLMs’ solution was evaluated by the author using Shariah scholar response as a reference point for authenticity. Findings ChatGPT and Gemini scored 70% in identification of Shariah compliance issues, providing Shariah basis/references which show proficiency. However, they scored relatively low 50% against third benchmark demonstrating limitations in offering Shariah guidance. Meta AI, scored 60% which shows relatively good performance only in referencing Shariah principles (60%) but struggled in issue identification and completely failed to provide Shariah guidance 40% and 0% respectively. Conclusion This study is conducted to investigate the capacity of Large Language Models (LLMs) in providing Shariah advisory services in complex Islamic financing scenarios. The results reveals that LLMs like ChatGPT and Gemini can be proved as valuable tools in initial evaluation of Islamic banking transactions for identification and referencing from Shariah point of view of issues. However, their ability to offer practical guidance from Shariah perspective on executed transactions is limited.

Document Type

Restricted Access

Submission Type

Research Project

Download

The full text of this document is only accessible to authorized users.

Share

COinS