AI in Credit Scoring
IOSR Journal of Economics and Finance · 2026-03 · 17 (3) · 67-76
DOI: 10.9790/5933-1703016776 ↗Kurz gefasst
Fachartikel in der Zeitschrift IOSR Journal of Economics and Finance, 2026-03. Verfasst von Aniketh Bharathwaj.
Abstract (Originalsprache)
The global financial system passes through the transformative restructuring under the pressure of the association of Artificial Intelligence (AI) and Machine Learning (ML) with the primary processes, which is mostly concerning credit rating and risk assessment. In this sector, the paper critically examines the extensiveness, complex nature of Al by providing it with a systematic review of its technological advances, social economic impacts of the technology and the governance challenges that have been raised with the technology. The reliance on the traditional credit scoring schemes which are simply rooted on the old statistical methodology and static, conventional data also have its own inherent weakness particularly on their capacity to effectively judge and include people with thin or no formal credit record. In their turn, al-based models leverage the computing capabilities of complex algorithms (e.g., Gradient Boosting Machines (GBM), Random Forests, and Deep Neural Networks (DNNs)) to leverage large-scale and granular alternative data sources (e.g., utility payments, mobile usage patterns, records of digital transactions, etc.). The combination of the empirical evidence provided within the framework of this paper is indicative of the fact that Al provides a verifiable and meaningful increment in the level of predictive accuracy, which is commonly 15 or 25 percent of the measures of predicting defaults relative to the traditional models. What is more important is the fact that this heightened accuracy is an efficient catalyst of financial inclusion that results in the responsible proliferation of credit services to hitherto marginalized and underserved populations of people across various economies. However, significant, compound issues are also present in the same revolutionary move, and they are largely founded on the issue that black-box models have never been entirely transparent, the opportunities of the reproduction and intensification of the past prejudices of history through artificial intelligence, and the paramount need to synchronize regulation. The paper introduces and elaborates Responsible Al Credit Scoring (RAICS) Framework- a conceptual framework that assists financial institutions and regulators in reaching ethical and sustainable Al implementation. The framework will also entail incorporation of Explainable AI (XAI) approaches, stringent fairness-aware machine learning optimization and a structurally sound and sustained model governance to ensure that the quest towards higher predictive accuracy will not result in the erosion of the same notion of fairness, transparency, and accountability in the highly-regulated credit market. It is the author of this paper that anticipates that there should be a cautious needed equilibrium to give priority to the technological innovation and high ethical protection levels to realize the successful long-term application of Al in credit scoring.
Zitieren
APA
Bharathwaj, A. (2026). AI in Credit Scoring. IOSR Journal of Economics and Finance, 17(3), 67-76. https://doi.org/10.9790/5933-1703016776
Vancouver
Bharathwaj A. AI in Credit Scoring. IOSR Journal of Economics and Finance. 2026;17(3):67-76. doi:10.9790/5933-1703016776
BibTeX
@article{Bharathwaj2026,
author = {Bharathwaj, Aniketh},
title = {AI in Credit Scoring},
journal = {IOSR Journal of Economics and Finance},
year = {2026},
volume = {17},
number = {3},
pages = {67--76},
doi = {10.9790/5933-1703016776}
}
EUVIMED