AI-Driven credit scoring and risk assessment in banks: Trends, opportunities, and challenges
Year: 2025
Type: article
Abstract: This paper analyzes the transformational shift in the banking sector from traditional credit risk assessment methods to advanced models driven by Artificial Intelligence (AI) and Machine Learning (ML). Conventional statistical models, such as logistic regression, are increasingly recognized as inadequate; they are often static, rely on narrow historical datasets, and systematically exclude "thin-file" or unbanked populations, particularly in developing economies. This review synthesizes the evolution of credit scoring, identifying three key technological trends redefining risk assessment: 1) the integration of alternative data sources (such as bank transaction data, utility payments, and digital footprints) to achieve financial inclusion; 2) the transition from static snapshots to real-time, dynamic scoring for proactive risk management; and 3) the advent of hyper-personalization in designing credit products. This shift presents a critical duality. On one hand, AI offers significant opportunities, including enhanced predictive accuracy, improved profitability, and the ability to extend formal credit to previously underserved populations. On the other hand, it introduces profound challenges, most notably the risk of amplifying systemic algorithmic bias, the complexities of regulatory compliance (such as the GDPR’s "right to explanation"), and the inherent opacity of the "black box" problem.
Source: The International tax journal
Authors: Halder, Sangjukta, Deshmukh, Renuka
Institution: MIT World Peace University
Language: English
Subfield: Accounting
Domain: Social Sciences
SDG: Reduced inequalities
Open Access status: diamond
OpenAlex ID: W7215591139
