摘要
符合可理解、可靠和可控等内在要求的可信算法是防范算法风险和构建算法信任的基础,算法的可信控制应该成为算法治理的核心。通过法律的可信控制,即按照法律嵌入和法律调节的规制逻辑,通过贯穿于算法的技术层、应用层和治理层的系统化制度安排,确保算法以可信任状态得以设计、部署、应用和执行,是实现算法可信的有效方式。系统地构建算法可信控制的制度体系,需要从算法本体的维度,通过完善透明度规则、推动算法伦理法律化和探索算法验证评估制度等措施确保算法自身的可信度,并从算法关系的维度,通过明确算法权利、强化算法问责和拓展监管体系等制度,约束算法相关主体的行为。
Trusted algorithms that meet the inherent requirements of comprehensibility,reliability,and controllability are the basis for preventing algorithm risks and building algorithm trust.Control on the trustworthiness of algorithms should be the core of algorithm governance.In accordance with the control logic of legal embedding and legal adjustment,through the systematic-institutional arrangement that runs through the algorithm's technical layer,application layer and governance layer,it is ensured that the algorithm can be designed,deployed,applied and executed in a trustworthy state,which is the effective approach to trusted control of the algorithm.From the dimension of the algorithm ontology,ensuring the credibility of the algorithm itself through transparency rules,algorithmic ethics legalization and algorithm verification and evaluation system,and from the dimension of the algorithm relationship,through the algorithm rights,algorithm accountability and regulatory framework and other systems to restrict the behavior of algorithm-related subjects,we can constitute a specific path to the legal control on the trustworthiness of algorithms,which is also a necessary institutional preparation before the arrival of the al-gorithm society.
出处
《东方法学》
CSSCI
北大核心
2021年第3期5-21,共17页
Oriental Law
基金
国家社科基金青年项目“系统重要性金融科技公司的法律规制”(项目批准号:20CFX062)的阶段性研究成果。
关键词
可信算法
算法治理
算法本体
算法关系
制度理性
技术理性
trusted algorithms
algorithm governance
algorithm ontology
algorithm relationship
institutional rationality
technology rationality
