财经问题研究

人工智能驱动的算法技术在带来经济效益的同时，也暗含算法歧视风险，因而人工智能背景下实现算法歧视的法治化治理成为时代所需。从产生根源看，算法歧视源于个性化消费需求、个性化定价、对网络信息的依赖、人工智能算法的“黑箱”特性及机体哲学视角下人机关系的变化，存在于互联网平台的多种场景之中，对个体权益、公众权益和市场竞争利益造成了侵害。对此，本文分析了目前中国算法歧视的规制现状及现实困境，在总结并借鉴欧盟和美国人工智能算法歧视治理实践经验的基础上，提出中国应构建算法歧视法治化治理体系，以算法透明度指标破解算法“黑箱”困境、以主流价值观引导算法“向善”发展、以完整行为认定标准应对算法歧视规制场景、以分层审查体系丰富算法歧视规制资源、以算法纠纷解决机制弥补消费者权益损失，逐步构建中国互联网平台算法歧视法治化治理框架，形成推动互联网平台和人工智能算法应用高质量发展的良性格局。

With the rapid advancement of artificial intelligence(AI) technologies, algorithmic discrimination has become increasingly prominent, posing a significant challenge for the governance of digital society. Existing studies have largely focused either on technological ethics or fragmented legal rules, while relatively few studies have examined algorithmic discrimination systematically from the perspective of legal institutional construction.Drawing on interdisciplinary theoretical perspectives, the paper first unpacks the underlying causes of algorithmic discrimination. The drive toward personalized consumption and dynamic pricing mechanisms accelerates differential treatment of users; heavy reliance on online information environments amplifies pre-existing data biases; the “black box” nature of algorithms conceals crucial aspects of decision-making; and shifts in the philosophical understanding of human-machine relations complicate the attribution of responsibility. On this basis, three typical application scenarios are identified: First, discriminatory pricing and algorithmic collusion. Second, discriminatory information recommendation and filtering. Third, biased automated decision-making. Methodologically, the paper employs a combination of normative analysis and comparative legal research. It systematically reviews China's existing regulatory framework. While these instruments provide an initial basis for oversight, the paper identifies four major dilemmas that hinder their effectiveness in addressing algorithmic discrimination. To respond to these challenges, this paper compares two dominant international models: the EU's rights-centered and data-protection-oriented regulatory approach, and the US' industry-led framework emphasizing corporate responsibility and self-regulation. The contribution of this paper lies in its departure from narrow, technology-centric or sector-specific perspectives, and in its development of a systematic governance framework that integrates theoretical analysis, problem identification, institutional critique, and path design. This holistic approach provides a reference for regulatory authorities in developing targeted and practicable tools, enabling them to simultaneously encourage algorithmic innovation and effectively guard against risks of discrimination. Furthermore, the framework contributes to broader scholarly debates by supplying analytical foundations for research on algorithmic accountability, AI ethics, and digital justice, while facilitating interdisciplinary dialogue and advancing theoretical integration in these frontier domains.