迈向信任的算法个性化推荐——“一键关闭”的法律反思  

Trust-Based Algorithmic Personalized Recommendations:Reflections on the Law of"One-Click-Off"

作　　者：林嘉[1] 罗寰昕 LIN Jia;LUO Huan-xin(Law School,Renmin University of China,Beijing 100872,China)

机构地区：[1]中国人民大学法学院,北京100872

出　　处：《编辑之友》2024年第3期79-88,共10页Editorial Friend

基　　金：国家社会科学基金重大项目“我国社会法的概念、原则、理论与实践”(18ZDA140);中国人民大学科学研究基金(中央高校基本科研业务费专项资金)资助项目(22XNH016)。

摘　　要：个性化推荐主要面临威胁人类主体自主性、“信息茧房”与“回音室”效应、大数据“杀熟”方面的质疑。“一键关闭”功能因能够实现隐私个人控制、促进信息自由流通和预防价格歧视而被推崇。然而,用户往往缺少隐私自治能力,个性化推荐造成群体极化之前首先具有议程凝聚、促进交往的价值,以关闭个性化推荐来解决价格歧视存在目标和手段的错位。“一键关闭”不仅难以符合规制预期,其自身还存在可操作度低和所承载公共价值模糊的弊端。个性化推荐的治理应以算法向善为方向,通过打造以信任为中心的透明可解释、用户交互友好的算法推荐,发挥个性化推荐的公共价值,增进社会福利。Personalized recommendations have been challenged as a threat to human autonomy,the"information cocoon"and"echo chamber"effect as well as the"killing of familiarity"by big data.The"one-click-off"function is promoted as a way to achieve personal control over privacy,promote the free flow of information and prevent price discrimination.However,users often lack the capacity for privacy autonomy,and algorithmic recommendations have the value of cohesive agendas and facilitating interactions before causing group polarization.In addition,there is a misalignment of goals and means to solve price discrimination by turning off personalized recommendations.The"one-click-off"option does not only fail to fulfill the functional expectations of legislations,but also suffers from poor operability and the ambiguity of the public values it carries.The reform of personalized recommendations should be oriented towards algorithmic goodness,and through the creation of trust-centered,transparent and interpretable algorithmic recommendations that are user-friendly.In the meanwhile,the public value of personalized recommendations should be brought into play to enhance social welfare.

关 键 词：个性化推荐 算法规制 可信算法 隐私个人控制 “信息茧房” 

分 类 号：G206[文化科学—传播学]