人工智能算法侵害的立法规制机制研究

Research on the Legislative Mechanism of Artificial Intelligence Algorithm Infringement

摘要: 人工智能算法是社会生产力发展和科技进步的产物，其广泛应用极大地促进了社会生产力水平的跃升与生产关系的变革，但也催生了一系列新型风险，包括个体层面的算法歧视和算法偏见，市场层面的算法共谋与算法垄断，国家层面的算法黑箱、算法霸权等，亟待采取立法规制。为破解当前监管机制不完善，算法责任界定不明等诸多局限，需要实现人工智能算法可解释、可评估、可监管等可信特征，有效防范人工智能算法风险。从人工智能算法侵害风险类型化的角度出发，提出对与人工智能算法发展有关的传统的法律制度进行适用性改进，并从立法方法、立法模式、立法路线三个方面提出规制人工智能算法的法律路径。

Abstract: Artificial intelligence algorithm is the product of the development of social productivity and scientific and technological progress. Its wide application has greatly promoted the improvement of social productivity and the transformation of production relations, but it has also spawned a series of new risks, including algorithm discrimination and algorithm bias at the individual level, algorithm collusion and algorithm monopoly at the market level, algorithm black box and algorithm hegemony at the national level, etc. Legislative regulation is urgently needed. In order to solve the current many other limitations such as the imperfect regulatory mechanism and the unclear definition of the algorithm responsibility, it is necessary to realize credible characteristics of artificial intelligence algorithm that can be explained, evaluated and regulated, effectively prevent the risks of artificial intelligence algorithm. From the perspective of AI algorithm’s infringement risk type, this paper proposes to improve the applicability of the traditional legal system related to the development of AI algorithm, and proposes the legal path to regulate AI algorithm from three aspects: legislative method, legislative model and legislative route.