算法影响评价:算法规制的制度创新  

Algorithm Impact Assessments:Institutional Innovation of Algorithm Regulation

作　　者：李安[1] Li An(Research Centre of Intellectual Property, Zhongnan University of Economics and Law, Wuhan 430073)

机构地区：[1]中南财经政法大学知识产权研究中心,武汉430073

出　　处：《情报杂志》2021年第3期146-152,161,共8页Journal of Intelligence

基　　金：高等学校学科创新引智计划(111计划)“新时代科技革命与知识产权学科创新”(编号:B18058);中南财经政法大学中央高校基本科研业务费专项资金资助项目“机器学习作品的著作权法研究”(编号:202011501)的研究成果之一。

摘　　要：[目的/意义]算法在推动智能化革命的同时,带了一系列包括算法黑箱、算法偏见在内的技术风险和社会问题,因此有必要从制度层面探讨算法规制的有效路径。[方法/过程]算法规制的要点是提高算法透明度。在与算法透明现有方法的比较分析中,提出算法影响评价的制度优越性。进而,从制度渊源和理论基础两个方面,对算法影响评价制度进行深刻剖析;并从分类管理、评价范围、程序环节、报告书等方面,对算法影响评价制度的具体构建进行系统讨论。[结果/结论]算法影响评价制度有助于构建一个取向于公开和公正的智能社会。中国应密切跟踪欧美国家(地区)的立法政策动向,采用“两步走”方案推进算法影响评价制度的构建和发展。[Purpose/Significance]While promoting the intelligent revolution,algorithms bring a series of technical risks and social problems,mainly including algorithm black box and algorithm bias.Therefore,it is necessary to explore the effective path of algorithm regulation from the institutional level.[Method/Process]The key point of algorithm regulation is to improve the transparency.In comparison with the existing methods of algorithm transparency,the institutional superiority of Algorithm Impact Assessments(AIAs)is proposed.Furthermore,AIAs is deeply analyzed from two perspectives called institutional origins and theoretical basis.And the specific construction of AIAs is discussed systematically from the aspects of classification management,the scope of assessment,the procedure of assessment,impact statement,etc.[Result/Conclusion]AIAs can contribute to building a transparent and fair algorithm society.China should closely keep pace with the legislative policy trend of European and American countries(regions),and take the"two-step"scheme to promote the construction and improvement of the AIAs.

关 键 词：人工智能 算法透明 算法影响评价 元规制 信息法学 

分 类 号：D923.49[政治法律—民商法学]