Li Yanan
With the development of artificial intelligence technology, artificial intelligence products are gradually integrated into social life, leading to a rise in product accidents. The relevant provisions of the Civil Code of China and the Product Quality Law of China are formulated based on traditional industrial products and fail to adequately address the special characteristics of artifi cial intelligence products. Due to the characteristics of artificial intelligence such as non-transparency and autonomous learning, it is determined that defect identification cannot be directly adopted in the current centralized and uniform manner. The diversifi cation of risks associated with artifi cial intelligence products requires that defect identification be taken in a decomposed manner, based on the sources of risk to determine the existence of defects. At the same time, the special characteristics of artifi cial intelligence products require the addition of new identifi cation standards. Finally, the specifi c identifi cation rules for artifi cial intelligence product defects should be constructed from four aspects: manufacturing defects, design defects, warning defects, and tracking and observation defects.
内容提要：随着人工智能技术的发展，人工智能产品逐渐融入社会生活，产品事故 的发生越来越频繁。我国《民法典》和《产品质量法》的相关规定主要是以传统工业产 品为蓝本制定的，难以适应人工智能产品的特殊性。由于人工智能具有不透明性和自主 学习等特征，这决定了缺陷认定不能单纯依赖目前的集中统一方式。人工智能产品的风 险来源复杂多样，因此缺陷认定应采取分解化方式，根据具体风险来源确定缺陷的存在。 同时，人工智能产品的特殊性要求补充新的认定标准。最后，需要从制造缺陷、设计缺 陷、警示缺陷和跟踪观察缺陷四个方面构建人工智能产品缺陷的具体规则。