The confluence of artificial intelligence and credit risk assessment-A scientometric analysis and research frontiers

The integration of artificial intelligence (AI) into credit risk assessment has evolved into a prominent and rapidly expanding field of research. Despite its growing importance, a comprehensive scientometric investigation capturing the intellectual landscape and thematic progression of this domain remains largely unexplored. This study addresses a key research gap by categorizing AI-assisted credit risk assessment into consumer and corporate domains—further distinguishing consumer credit into bank and P2P lending—and mapping major methodological stages often overlooked in prior scientometric analyses. The present study conducts an extensive scientometric analysis of 2151 peer-reviewed articles published between 2000 and 2025, sourced from the Web of Science (WoS) database. This review presents a longitudinal analysis of leading country- and discipline-specific contributors, thematic categorization, co-citation structures, and collaborative networks that shape the integration of AI in credit risk assessment. The analysis highlights China (having 28% publications) as a leading contributor, with the South Western University of Finance and Economics (5% publications) emerging as a key collaborating institute. Data preprocessing was explicitly reported in only 5.9% of studies, with class imbalance mitigation addressed in 62% of these, predominantly via SMOTE. Feature preprocessing appeared in 19% of publications, chiefly through feature selection, with LASSO as the leading technique. Ensemble methods for data modeling were mentioned in 36% of the studies (2020–2025), most commonly employing Random Forest. A shift in publication trends is observed—from applying AI for classifying bad debts to addressing ethical concerns such as accountability, fairness, data protection driven by evolving regulations like GDPR, PIPL, and others. The paper concludes by identifying future research priorities, including IoT-driven and image-based unstructured text analytics for real-time risk assessment, sentiment scoring in P2P lending, and the integration of sustainability metrics into corporate bankruptcy prediction, alongside strengthened ethical AI frameworks.