Across three pre-registered experiments and one pre-registered field survey study, we demonstrate that benevolence is an important, but previously neglected, predictor of trust in algorithmic management. We demonstrate that perceived benevolence of algorithmic managers is significantly lower than that of human managers, which harms trust in algorithmic management, and that these effects are robust when we control perceived ability and integrity. We link AI management’s lack of emotions to perceived benevolence and trust and find that in a context where individuals want empathy, AI’s lack of emotions will be particularly salient and AI management will be perceived as less benevolent and trustworthy, compared to human management. Combined, these findings deepen our understanding of trust...[ Read more ]

Across three pre-registered experiments and one pre-registered field survey study, we demonstrate that benevolence is an important, but previously neglected, predictor of trust in algorithmic management. We demonstrate that perceived benevolence of algorithmic managers is significantly lower than that of human managers, which harms trust in algorithmic management, and that these effects are robust when we control perceived ability and integrity. We link AI management’s lack of emotions to perceived benevolence and trust and find that in a context where individuals want empathy, AI’s lack of emotions will be particularly salient and AI management will be perceived as less benevolent and trustworthy, compared to human management. Combined, these findings deepen our understanding of trust and provide practical advice on the adoption of algorithmic management.

Keywords: Trust; Benevolence; Emotions; Algorithmic Management