Employees Adhere More to Unethical Instructions from Human Than AI Supervisors: Complementing Experimental Evidence with Machine Learning

Abstract

The role of artificial intelligence (AI) in organizations has fundamentally changed from performing routine tasks to supervising human employees. While prior studies focused on normative perceptions of such AI supervisors, employees’ behavioral reactions towards them remained largely unexplored. We draw from theories on AI aversion and appreciation to tackle the ambiguity within this field and investigate if and why employees might adhere to unethical instructions either from a human or an AI supervisor. In addition, we identify employee characteristics affecting this relationship. To inform this debate, we conducted four experiments (total N = 1701) and used two state-of-the-art machine learning algorithms (causal forest and transformers). We consistently find that employees adhere less to unethical instructions from an AI than a human supervisor. Further, individual characteristics such as the tendency to comply without dissent or age constitute important boundary conditions. In addition, Study 1 identified that the perceived mind of the supervisors serves as an explanatory mechanism. We generate further insights on this mediator via experimental manipulations in two pre-registered studies by manipulating mind between two AI (Study 2) and two human supervisors (Study 3). In (pre-registered) Study 4, we replicate the resistance to unethical instructions from AI supervisors in an incentivized experimental setting. Our research generates insights into the ‘black box’ of human behavior toward AI supervisors, particularly in the moral domain, and showcases how organizational researchers can use machine learning methods as powerful tools to complement experimental research for the generation of more fine-grained insights.

The Now, New, and Next of Digital Leadership: How Artificial Intelligence (AI) Will Take Over and Change Leadership as We Know It

“…Relatedly, third, while initial evidence indicates that humans are less likely to follow the unethical instructions of an AI (versus human) leader (Lanz et al, 2023), at the same time humans seem to experience less moral outrage over algorithmic discrimination than over human discrimination (Bigmann et al, 2022). Accordingly, it contains a risk that decisions of AI leaders that one “has to follow” are used as an excuse for unethical behavior with implications for the potential weakening of collective action to address systematic discrimination and other societal issues (Bigmann et al, 2022).…”

“…In this way, we can hopefully maintain a “human-in-the-loop” pattern (Grønsund & Aanestad, 2020) whereby human leaders still (co-)generate a ground truth against which to assess algorithmic leadership and potentially adapt the underlying AI. Students need to develop a digital backbone in order to stand their ground; against the technology itself when it provides ethically questionable advice (e.g., firing certain employee groups because they underperform, Lanz et al, 2023); against engineers who only see the opportunities of the machine (Köbis et al, 2021); and against a multitude of consultants who want to integrate ever new technologies without considering their broader impact.…”

A Literature Review of Human–AI Synergy in Decision Making: From the Perspective of Affordance Actualization Theory

“…The first stream focus on the roles of humans-AI and task division during the collaboration. In collaborative decision-making between humans and machines, AI can assume different roles, such as a facilitator, reviewer, expert advisor, or guide [5,15]. With the improvement of AI automation, agents' roles have evolved from being auxiliary tools to active participants in team decision-making processes [16], leading to a wider array of team collaboration formats.…”

“…Organizational Outcomes [1] hiring and firing employees algorithm-enabled software system humans' acceptance of machine participation [7] service encounter intelligent digital voice assistant users' motivations to adopt AI [15] human resource management algorithm-based AI system human reaction to an AI supervisor [18] welcoming visitors and employees and offering directions to specific locations on a campus humanoid robots trust, intention to use, and enjoyment [19] clinical decision-making on Rehabilitation Assessment AI-based decision-support-system usefulness and attitudes toward the system [20] socially shared regulation (SSRL) in learning intelligent agent learning regulation improvement [26] Scout Exploration Game explainable artificial intelligence (XAI) system aligned understanding between humans and AI [27] reconnaissance missions to gather intelligence in a foreign town algorithm-based robots transparency, trust, mission success, and team performance [28] table-clearing task autonomous-system-based robot assistants human-robot team performance and trust evolution [29] real-time human-robot cooking task algorithm-based XAI collaboration performance and user perception of the robot [30] make a cup of coffee or clean the bathroom a synthetic robot maid more explainable robot behavior, team performance [31] a human-robot team preparing meals in a kitchen autonomous-system-based robot assistants human-robot collaboration performance [32] turn this plate of food into a low-carb meal recommend system (XAI) team performance [33] nutrition-related decision-making task recommend system (XAI) objective performance, trust, preference, mental demand, and understanding [34] deception-detection task machine learning models human performance and human agency [35] deception-detection task machine learning models human performance [36] multi-label image classification machine learning algorithms outcome prediction accuracy and confidence [37] three high-stakes classification tasks machine learning algorithm performance/compatibility tradeoff [38] recruitment and staffing, e-commerce, banking AI-based platform/assistant degree of fairness, transparent feedback, less-biased decisions [39] managerial decision-making such as hiring and firing employees algorithm-based AI system acceptance of the decisions [40] open medicine bottles autonomous-system-based robot human trust in the robot [41] naming and distinguishing species machine learning classifier end users' appropriate tru...…”

“…AI can participate in team collaboration as different roles (leader/participant) and improve the efficiency of team performance to some extent [15,92,98]. However, in the practice of human-AI team collaboration, the lack of explainability of AI algorithms or AI systems remains to be addressed, which is mainly caused by the "black box" of AI technology and algorithm.…”