首页|政府算法决策风险的生成机理与防范策略——基于符号互动论的分析框架

政府算法决策风险的生成机理与防范策略——基于符号互动论的分析框架

扫码查看
政府算法决策风险是人工智能算法嵌入政府决策过程中所形成的新风险类型。基于符号互动论的视角,构建符号"主体—媒介"的分析框架,可以发现政府算法决策风险实质是政府算法决策相关主体在互动过程中对符号媒介价值的整体性偏离,易导致政府决策产生不能达到其目标的可能性与不确定性。技术、主体、制度层面的主客观抵牾与隐患是引致政府算法决策风险生成的重要诱因,其生成过程遵循决策"互动价值偏离—互动媒介缺陷—互动场域差异—人机互动阻滞"的生成链条。为更好地规避政府算法决策风险,可采取强化算法技术的公共性价值、提升决策相关主体的数字素养、完善政府算法决策的制度机制等适数化策略,提升政府治理效能。
The Generation Mechanism and Prevention Strategy of Government Algorithm Decision-Making Risk:The Analytical Framework Based on the Symbolic Interaction Theory
Government algorithm decision-making risk is a new type of risk formed when artificial intelligence algorithm is embedded in government decision-making process.Based on the symbolic interaction theory,the"subject-medium"analytical framework is constructed.It can be found that the government algorithm decision-making risk is the overall deviation of the value of symbol media in the interaction process of relevant subjects of government algorithm decision-making,resulting in the possibility and uncertainty of government decision-making that cannot reach its goal.Subjective and objective contradictions,as well as hidden dangers at the levels of technology,subject and system are important incentives for the generation of government algorithm decision-making risks,and the generation process follows the decision-making chain of"interactive value deviation,interactive media defects,interaction field difference,and human-computer interaction block".In order to better avoid the government algorithm decision-making risk,appropriate strategies such as strengthening the public value of algorithm technology,improving the digital literacy of decision-related subjects,and improving the institutional mechanism of government algorithm decision-making can be adopted to improve the efficiency of government governance.

government algorithm decision-making riskfitnesssymbolic interaction theory

周济南、苏厚任

展开 >

中共河南省委党校公共管理教研部

湘潭大学公共管理学院

政府算法决策风险 机理 符号互动论

2025

理论月刊
湖北省社会科学联合会

理论月刊

北大核心
影响因子:0.565
ISSN:1004-0544
年,卷(期):2025.(1)