Decoding digital signals: AI sentiment and financial performance at İslamic banks

Abstract

This study provides empirical evidence on the role of artificial intelligence (AI) and machine learning (ML) sentiment in influencing financial performance by Islamic banks. Using advanced textual analysis methods, including long short-term memory (LSTM) networks, AI and ML sentiment is derived from annual reports. The study employs fixed-effects regression using the return on equity (ROE) as the primary measure and robustness checks using random forest models and spline regressions to examine their impact on ROE and the return on assets (ROA). It also investigates the mediating role of the development of information communication technologies (ICT) and the moderating effect of growth in the gross domestic product (GDP). Our findings reveal that positive sentiment about AI and ML significantly enhances profitability, and combined sentiment has the strongest predictive power. The mediating role of ICT development highlights the importance of digital infrastructure, and GDP growth emphasizes the contextual dependence of AI-driven innovation.