以“智能向善”推进算法推荐的乱象治理  

Promote Chaos Governance in Algorithmic Recommendation with“AI for Good”

作　　者：王林生[1] WANG Linsheng(Center for Cultural Study,Beijing Academy of Social Sciences,Beijing 100101,China)

机构地区：[1]北京市社会科学院文化研究所,北京100101

出　　处：《治理现代化研究》2024年第2期71-77,共7页Governance Modernization Studies

基　　金：国家社会科学基金重点项目“十八大以来党领导文化建设的原创性实践和经验研究”(22AZD061)。

摘　　要：算法是互联网时代的核心技术支撑,但在具体应用情境中出现了信息定向推送造成的“茧房”效应、算法应用平台过度追求效率导致的操控效应、用户路径依赖带来的“杀熟”效应等算法推荐乱象。对算法推荐不良行为的自我辩护,坚持的是一种技术“工具—独立”的价值观,遵循唯技术主义,但脱离了具体的情境应用。推进算法推荐乱象的治理和行为纠偏,应在算法推荐应用情境中构建以“智能向善”为价值导向的治理体系。这需要从源头设计层面赋予算法编码以正向的价值导向,从运行应用层面强化算法应用平台的主体责任建构,从政策保障层面厘清算法应用平台的利益诉求边界,以及从监管实施层面提升监管部门甄别处理流量事件的应急能力。Algorithms are the core technological support in the Internet era.In the specific application context of algorithms,there have been problems such as the cocoon effect caused by information directional push,the manipulation effect arising from the excessive pursuit of efficiency by algorithm recommendation platform,and the killing effect arising from user path dependence.These problems are caused by the targeted push of algo-rithms.When algorithms defend themselves for bad behavior,they adhere to a technological“tool-indepen-dent”value orientation,follow the technologism,and separate from specific contextual applications.To promote the governance of the chaos of algorithm recommendation,we should build a governance system with the value orientation of“AI for Good”in the application context of algorithm recommendation.This requires giving posi-tive value orientation to algorithm coding from the source design level,strengthening the construction of the main responsibility of the algorithm recommendation platform from the operation application level,clarifying the boundary of the interest demands of the algorithm recommendation platform from the policy guarantee level,and enhancing the emergency ability of regulatory authorities to identify and respond to traffic events from the supervision implementation level.

关 键 词：算法 行为失范 情境 智能向善 治理 

分 类 号：C913.4[经济管理]