CREDIT DECISION AUTOMATION IN COMMERCIAL BANKS: A REVIEW OF AI AND PREDICTIVE ANALYTICS IN LOAN ASSESSMENT

Abstract

The increasing integration of artificial intelligence (AI) and predictive analytics in commercial banking has fundamentally transformed credit decision-making, enabling faster, more accurate, and more inclusive loan assessment processes. This systematic review aims to synthesize the current academic and empirical literature on AI-powered credit decision automation, with particular attention to methodological advancements, operational efficiency, financial inclusion, ethical governance, and regulatory challenges. Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, a total of 102 peer-reviewed studies published between 2000 and 2023 were selected and analyzed from major databases including Scopus, Web of Science, IEEE Xplore, ScienceDirect, and Google Scholar. The review finds that machine learning models particularly ensemble methods and deep neural networks consistently outperform traditional statistical approaches in credit scoring accuracy, especially in complex borrower environments. Operationally, AI-driven systems significantly reduce loan processing time and operating costs, while enabling real-time credit adjudication and scalability across diverse lending portfolios. Furthermore, the use of alternative data, such as mobile phone metadata, utility payments, and psychometric testing, has expanded credit access to previously underserved groups, demonstrating the potential of AI to promote financial inclusion. However, the review also identifies significant concerns around algorithmic bias, model transparency, and compliance with legal frameworks such as GDPR, ECOA, and FCRA. To address these issues, the literature increasingly supports the adoption of explainable AI (XAI) methods, fairness-aware algorithms, and ethics-by-design principles in model development and deployment. Overall, this review highlights that while AI and predictive analytics offer transformative potential in automating credit decisions, their effectiveness depends on the balance between technological sophistication, ethical responsibility, and regulatory alignment. The findings contribute a comprehensive foundation for future research, policy formulation, and strategic implementation of credit automation systems in the evolving landscape of digital finance.