面向人机协同的公共决策转型:内涵、议题与框架

Towards Human-AI Collaborative Public Decision Making:Concepts,Issues,and Framework

随着基于大数据、人工智能等技术的决策算法被越来越多地应用于公共决策,公共决策正在逐渐出现由传统经验决策向人机协同决策的转型。但如何将前沿算法技术有效嵌入决策过程、真正实现决策质量提升还需解析一系列人机协同决策中的基础机制。本文旨在阐释人机协同决策内涵,并跨学科梳理相关研究进展,为未来研究提供框架与参考。文章辨析了人机协同公共决策的概念内涵与主要过程,梳理当前研究关注的重点议题,包括个体层面的算法认知基础机制、多元主体层面的裁量权变化与影响以及算法参与公共决策的潜在风险,构建了由微观个体算法认知机制、中观多元主体-算法互动机制以及宏观综合影响三个层次研究内容所构成的人机协同公共决策分析框架,并提出了若干值得探索的未来研究方向。

The emergence of big data and progresses in artificial intelligence are driving the application of algorithms in public decision-making.Correspondingly,there is a modal shift in public decision-making towards human-AI collaborative decision-making.However,to truly integrate cutting-edge algorithms into public decisions and effectively enhance the quality of decisions,it is necessary to explore the fundamental mechanisms of how human and AI interact in public decision-making contexts.To fill the gap,this article aims to provide a review of relevant research from multiple disciplines and provide a roadmap for future research.The article starts from conceptualizing human-AI collaborative decision-making in the public sector,providing a conceptual model on the collaboration process.The key research topics include decision-makers’perception and attitudes towards algorithms at the individual level,the(re)distribution of discretion at the organizational level,and the potential risks of algorithm application in public-sector decision-making.Finally,the article proposes a research framework composed of algorithm perception at the micro-level,multiagent-algorithm interaction at the meso-level,and impact evaluation of this new decision-making approach at the macro-level.A number of future research questions are also proposed.