Explainable Artificial Intelligence for Financial Services: A Bibliometric Review

Abstract

Artificial intelligence has become an essential technology in the banking, financial services, and insurance (BFSI) sector, supporting tasks such as risk assessment, fraud detection, and financial forecasting. However, the increasing use of complex “black-box” models, including deep learning and generative AI, raises serious concerns related to transparency, trust, and regulatory compliance. This study aims to provide a clear overview of the research landscape on Explainable Artificial Intelligence (XAI) in financial services by identifying key trends, commonly used methods, and existing research gaps. We analyzed 580 publications indexed in the Scopus database from 2018 to 2024 using bibliometric techniques, supported by VOSviewer and Latent Dirichlet Allocation for thematic analysis. The results reveal a rapid growth in XAI-related publications, particularly after 2021, with post-hoc explanation methods such as SHAP and LIME being the most widely adopted. At the same time, our findings indicate that explainability remains limited for emerging generative models, including GANs and Transformer-based architectures, especially in applications like fraud detection and financial prediction. Overall, this paper provides practical insights and a strategic reference for researchers and practitioners seeking to align advanced AI models with the transparency and accountability requirements of the financial sector.