学术报告

Safety in Obedience: Employees’ Reluctance to Deviate from their Supervisors’ Advice and the Role of AI
发布时间:2026-05-27 浏览次数:10

主题:Safety in Obedience: EmployeesReluctance to Deviate from their Supervisors’ Advice and the Role of AI

题目:服从即安全:员工对偏离主管建议的抵触与人工智能的作用


时间:  20250616日 上午1000


地点: 管理学院第教室


主讲人:Huaxiang Yin,  Nanyang Technological University

主持人:黄锨

Bio:

殷华祥博士于2014荷兰蒂尔堡大学获得会计学博士学位加入南洋商学院,目前为南洋商学院会计系副教授。他的研究主要运用实验方法,探究会计信息如何在业绩评估、薪酬设计及管理控制系统中用于管理决策,特别关注公平感、他人偏好及社会规范等社会动机的影响。在近期研究中,他借助皮电活动测量、眼动追踪及脑部扫描等前沿技术,分析人工智能在管理中的应用如何影响上述场景下会计信息的使用。他的研究成果发表于The Accounting ReviewJournal of Accounting ResearchAccounting Organization and Society等顶级学术期刊。其博士论文荣获美国会计学会管理会计分会杰出论文奖。


照片:

 

Abstract: 

Companies invest heavily in artificial intelligence–powered advisory systems to support managers and employees in decision making, often with the expectation that these systems can substitute for supervisors’ advisory roles. This study examines how the source of advice—supervisor versus AI—affects employees’ utilization of advice. Drawing on psychology literature of self-serving attribution bias, we predict that employees are reluctant to deviate from supervisors’ advice because they anticipate negative supervisory reactions during subsequent performance evaluations, and that this reluctance is mitigated when advice is provided by an AI advisory system. We test our theory in an interactive laboratory experiment in which participants act as employees who decide on a production level and may request advice, and supervisors who provide advice upon request and subsequently determine employees’ rewards. We manipulate whether the advice source is the employee’s supervisor or an AI advisory system that replicates supervisors’ advice. Consistent with our predictions, employees deviate less from supervisors’ advice than from AI advice. Moreover, supervisors award lower bonuses to employees who deviate from their advice than to those who follow it, whereas this negative effect is muted when advice is provided by AI. Our findings contribute to the literature on advice taking, subjective performance evaluation, and the growing role of AI as an alternative source of advice in organizations.