摘要
[目的/意义]人智信任是实现高质量人智协作的基础与前提,人智协作的实践深化也进一步丰富了人智信任的视角特征。因此,在人智协作情境下解构人智信任的视角特征,对理解、巩固人智信任具有重要意义。[方法/过程]基于ANN和Q方法对人智信任进行维度划分与实证检验。一方面采用Q方法,从近10年来的相关文献对人智信任的形成视角进行定量聚类,对其特征构成进行定性归纳;另一方面采用ANN方法,基于ChatGPT的模拟情景实验对人智信任的视角特征进行实证检验。[结果/结论]Q方法的定量分析发现人智信任具有社会实体和作业工具的双重视角;Q方法的定性分析则进一步解构了两视角下人智信任的具体特征。具体而言,社会实体视角下的人智信任可以从感知、认知、评判维度解析,作业工具视角下的人智信任则由确定性、稳定性和机器能力的特征构成。此外,ANN的实证检验结果表明不同人智信任视角下的多重特征对用户满意度和持续使用倾向具有显著影响,即证明了对人智信任的维度划分具有较好的适用性。
[Purpose/significance]Human-AI trust is the basis and premise of high quality human-AI cooperation.The deepening of human-AI cooperation has further enriched the perspective characteristics of human-AI trust.Therefore,deconstructing the perspective features of human-AI trust in the context of human-AI cooperation is of great significance to understanding and consolidating human-AI trust.[Method/process]Based on ANN and Q methods,this study attempts to divide the dimensions of human-AI trust and empirically test it.On the one hand,based on the relevant literature in the past decade,the paper quantitatively clusters the formation perspective of human intelligence trust and qualitatively summarizes its characteristics.On the other hand,the simulated scenario experiment of ChatGPT is taken as an example to empirically test the perspective characteristics of human-AI trust through ANN.[Result/conclusion]Firstly,the quantitative analysis of Q method found that human-AI trust has dual perspectives of social entity and operational tool.The qualitative analysis further deconstructs the concrete characteristics of human intelligence trust from the two perspectives.Specifically,human-AI trust from the perspective of social entity can be analyzed from the dimensions of perception,cognition and judgment,while human-AI trust from the perspective of operational tools is composed of the characteristics of certainty,stability and machine ability.In addition,ANN’s empirical test results show that multiple features from different perspectives of human-AI trust have a significant impact on user satisfaction and continuous use behavior,which proves that the dimension division of human-AI trust in this study has good applicability.
作者
张进澳
卢新元
周一夫
孟华
王雪霖
Zhang Jin’ao;Lu Xinyuan;Zhou Yifu;Meng Hua;Wang Xuelin(School of Information Management,Central China Normal University,Hubei Wuhan 430079;E-Commerce Research Center of Hubei Province,Central China Normal University,Hubei Wuhan 430079)
出处
《情报理论与实践》
北大核心
2025年第5期128-137,共10页
Information Studies:Theory & Application
基金
国家社会科学基金重点项目“数智时代下AIGC服务模式及生态治理研究”的成果,项目编号:23AGL040。
