摘要
人工智能算法应用于个人信用评分行业,极大降低了消费信贷等活动中人为因素的影响,提高了放贷效率。但是人工智能算法的诸多固有缺陷,包括基础数据质量缺陷、算法内嵌歧视性变量、机器学习的不确定性等,加之运用过程中的算法滥用,极易破坏金融公平,使所谓技术中立只是为信用评估提供了“客观的表象”。人工智能算法主导下的信用评分系统中,数据与算法担负起交易主体关系建构与秩序塑造的基本功能,而传统信用立法已严重滞后于对金融公平的现实需求。人工智能信用评分系统对个人获得公平信贷机会等基本权益影响巨大,并左右着金融资源的合理公平分配。信用立法中应增加数据和算法公平性规则,包括强调信用数据相关性和代表性要求等数据质量规则,增加金融消费者等弱势群体对算法的知情权、拒绝权等。
The application of AI algorithms in the personal credit scoring industry has significantly reduced human influence in activities such as consumer credit lending and thereby enhanced operational efficiency. However, the inherent flaws of AI algorithms-including deficiencies in foundational data quality, discriminatory variables embedded within algorithmic models, and the unpredictability of machine learning-coupled with algorithmic misuse in practice, risk undermining financial fairness. This renders the notion of “technological neutrality” merely an illusion of objectivity in credit assessments. Within AI-driven credit scoring systems, data and algorithms assume the fundamental role of constructing transactional relationships and shaping order, while traditional credit legislation has lagged far behind the demands of ensuring financial equity. AI-powered credit scoring impacts individuals’ rights to fair access to credit and influences the equitable allocation of financial resources. To address these challenges, credit legislation should incorporate data and algorithmic fairness rules, including requirements for the relevance and representativeness of credit data to ensure quality, and adequate safeguards for financial consumers’ rights regarding algorithmic transparency and rights to reject algorithmic decisions.
出处
《东南大学学报(哲学社会科学版)》
北大核心
2025年第3期47-55,154,155,共11页
Journal of Southeast University(Philosophy and Social Science)
基金
国家社科基金一般项目“人工智能算法应用的金融法规制”(22BFX089)阶段性研究成果。
