The automation of leadership functions: Would people trust decision algorithms?
The automation of leadership functions: Would people trust decision algorithms?
复制标题
领导职能的自动化：人们会信任决策算法吗？
DOI：
10.1016/j.chb.2020.106635
复制
发表时间：
2021-01-07
影响因子：
9.9
通讯作者：
Hertel, Guido
中科院分区：
文献类型：
--
作者：
Hoddinghaus, Miriam;Sondern, Dominik;Hertel, Guido
基于算法的决策的日益成熟使计算机能够执行许多领导职能。然而，成功实施的一个核心先决条件应该是人类员工信任这种自动化的领导代理。本研究（N = 333名员工）采用实验小插图方法和假设的工作场景，比较了参与者对自动化和人类领导的反应。我们操纵了领导主体类型（人vs.计算机）和决策主体类型（纪律vs.指导），并测量了参与者对领导主体的可信度感知和信任程度。结果表明，参与者认为自动化领导代理比人类领导代理在诚信和透明度方面更高。然而，人类领导者被认为更具适应性和更仁慈。在感知数据处理能力或作为决策主体的函数方面没有发生差异。感知可信度预测了对领导代理人的信任，而信任反过来又与进一步的工作相关结果（如感知决策的公平性、感知组织支持）正相关，证实了信任与组织的一般相关性。研究结果有助于我们理解自动化领导中的信任，并为基于计算机的领导决策提供实际意义。
The advancing maturity of algorithm-based decision-making enables computers to perform many leadership functions today. However, a central precondition of successful implementation should be that human workers trust such automated leadership agents. The present study (N = 333 workers) compared participants' reactions towards automated and human leadership using experimental vignette methodology with hypothetical work scenarios. We manipulated type of leadership agent (human vs. computer) and decision subject (disciplinary vs. mentoring), and measured participants' trustworthiness perceptions and trust in the leadership agent. Results showed that participants perceived automated leadership agents as being higher on integrity and transparency than human leadership agents. However, human leadership agents were perceived as more adaptable and more benevolent. No differences occurred with respect to perceived data processing capacity or as a function of decision subject. Perceived trustworthiness predicted trust in the leadership agent, which in turn was positively related to further work-related outcomes (e.g., perceived fairness of the decision, perceived organizational support), confirming the general relevance of trust for organizations. The results contribute to our understanding of trust in automated leadership and offer practical implications for computer-based decision-making in the leadership context.