An Exploration of the Dual Impacts of Artificial Intelligence Credit Scoring on Banking System Risk

Abstract

The digital transformation of finance continues to deepen under the dual policy and practical drive of the implementation of the "Interim Measures for the Administration of Generative Artificial Intelligence Services" and the deepening promotion of inclusive finance. Artificial intelligence credit scoring has become the core technical support tool for commercial banks' risk management, and its impact on banking system risks presents a distinct dual nature. This article combines the practice of government-bank data collaboration with real-world commercial bank operations to systematically sort out the technical operating logic and risk action mechanism of artificial intelligence credit scoring: on the one hand, it analyzes its positive enabling value in mitigating traditional credit risks and expanding the coverage of inclusive financial services; on the other hand, it deeply analyzes new risk hazards induced by data collection bias, model "black box" characteristics, and cross-institutional systemic transmission of risks. The study found that after a joint-stock bank optimized its credit assessment model through multi-source heterogeneous data fusion, the non-performing loan loss rate dropped by 30 percentage points compared with before optimization; However, due to a lack of training data for rural customers and small and micro businesses, some banks experienced a misjudgment rate of up to 28% for normal transactions in rural areas. This article ultimately constructs a collaborative governance framework of "data governance—algorithm transparency—regulatory adaptation." This framework uses data governance to ensure source quality assessments, algorithm transparency to address model interpretation challenges, and regulatory adaptation to address technological iterations. This framework provides a practical path for balancing financial technology innovation and risk prevention and control, while also offering a reference for technology application by small and medium-sized banks and supporting the robust operation of the banking system.