Managerial duties are increasingly fulfilled by artificial intelligence (AI). This may cause difficulties in roles requiring high emotional engagement, according to HKUST’s Mingyu Li and T. Bradford Bitterly. Lacking empathy, AI is viewed as less benevolent and trustworthy than humans, highlighting the need for organizations to deploy it carefully.

As the AI revolution reaches the management sphere, more workers will find themselves being organized and evaluated partly—or even entirely—by algorithms. However, as Li and Bitterly point out, “research provides equivocal findings regarding trust in algorithmic management,” suggesting that workers may have mixed feelings about AI managers.

Employee acceptance of management, whether human or AI, depends on trust. This must be earned. “Supervisors are more likely to gain the trust of their subordinates,” note the researchers, “if they are perceived to have a high level of ability, benevolence, and integrity.” They hence investigated whether AI management was perceived to possess these traits and how this affected acceptance.

A survey of delivery riders found that those with AI management rated it as less benevolent (“kind,” “warm,” and “caring”) than hybrid human/AI management, resulting in lower trust. AI managers were also perceived as less capable. In a follow-up experiment involving a hypothetical scenario, perceived benevolence mediated the effect of management type on trust.

AI’s perceived low benevolence has important effects in situations demanding empathy. This was shown by an experiment where participants involved in an imagined traffic accident increasingly wished to switch away from AI management. In another scenario, having an AI manager was perceived as more problematic when requesting leave due to bereavement than merely for a vacation.

“Organizations,” the researchers conclude, “should anticipate trust and adoption issues when replacing roles and responsibilities historically occupied by humans with AI systems.” Their study provides crucial guidance for addressing these hurdles.