قياس الامتثال الشرعي عبر الذكاء الاصطناعي: دراسة مقارنة لأربع أدوات وفق معايير هيئة المحاسبة والمراجعة للمؤسسات المالية الإسلامية

Abstract

Purpose: The research evaluates the performance of four Artificial Intelligence (AI) tools complying with the Shariah standards issued by the Accounting and Auditing Organization for Islamic Financial Institutions (AAOIFI) The assessment is based on quantitative and qualitative indicators derived directly from the standards.

Methodology: The research adopts a comparative analytical approach, combining quantitative and qualitative dimensions. Six core indicators were developed: reference to the title of the standard, standard number, clause numbers, lexical congruence, substantive congruence, and clarity and internal consistency. These were applied to four AI tools: Compliance with Shariah Standards, Islamic Banking Advisor, Shariah Financial Advisor, and ChatGPT 4.0.

Findings: The results reveal significant variation across the tools. ChatGPT 4.0 ranked highest in overall performance, though it showed notable weaknesses in reference accuracy. The other tools performed relatively better in lexical or substantive congruence. The findings also suggest that relying exclusively on such models may lead to fiqh-related inconsistencies or the omission of essential requirements.

Originality: This research is the first systematic attempt to combine quantitative and qualitative analyses in evaluating AI compliance with AAOIFI Shariah standards. It addresses a key research gap in assessing AI systems within the framework of institutional Shariah governance.

Practical Implications: The findings highlight the need to develop domain-specific language models trained directly on AAOIFI Standards to improve reliability in Shariah rulings and Islamic financial advisory services.