算法透明何以提升公众的政府信任?  

Can Algorithmic Transparency Enhance Citizen Trust in Government?

作　　者：樊博[1] 李晶晶 Bo Fan;Jingjing Li

机构地区：[1]上海交通大学国际与公共事务学院、应急管理学院

出　　处：《公共行政评论》2024年第1期4-24,M0003,共22页Journal of Public Administration

基　　金：国家自然科学基金重点项目(72134005);国家自然科学基金面上项目(71974128)。

摘　　要：随着算法辅助决策逐渐嵌入公共部门,算法黑箱极易冲击公众对于政府公权力的信任。算法透明成为政府治理的应有之义。然而,算法透明何以实现,以及政府信任与算法透明的关联仍存在争议。论文基于“金税工程”税务算法现实案例,设计了一个情景调查实验,实证检验了算法透明的实现机制及其对政府信任的影响路径。研究发现:(1)算法过程透明、算法结果透明均对公众感知算法透明具有显著正向影响;(2)感知算法透明能够提高公众对政府的信任;(3)算法信任在感知算法透明、政府信任中发挥了部分中介效应。研究明晰了算法透明的概念,回应了算法透明与信任的迷思,并阐明了其中的强化机制,为推进算法善治提供了实践参考。As algorithms become embedded in the public sector,the resulting black box of decisionmaking could threaten the public's trust in authority.Transparency has become an essential pursuit of modern governance,but trust in government as it relates to algorithmic transparency is still controversial.Based on the algorithms that are part of China's Golden Tax Project,this study used a vignette survey experiment to empirically examine the pathways of algorithmic transparency and its impact on citizen trust in government.Our findings suggest that:(1)processbased algorithmic transparency and resultbased algorithmic transparency both have positive effects on the citizens'perceived algorithmic transparency;(2)perceived algorithmic transparency can improve citizen trust in government;and,(3)citizen trust in algorithms plays a partial mediating effect in perceived algorithmic transparency and citizen trust in government.This paper clarifies the concept of algorithmic transparency,empirically responds to the myth of algorithmic transparency and trust,elucidates the strengthening mechanism therein,and suggests practical implications for promoting good governance in the algorithm era.

关 键 词：算法透明 政府信任 算法信任 调查实验 

分 类 号：D63[政治法律—政治学]