Using artificial intelligence at work has deeper effects than making jobs easier and increasing productivity. It elicits a complicated emotional response in employees that makes them want to spend more time with colleagues, but also causes them to feel lonelier, suffer from insomnia, and drink more, a March study found.
The findings show the psychological impact of introducing A.I. in the workplace that companies will increasingly have to consider as they add the technology to the normal flow of business. A.I. may be just a tool, but it comes with a lot of baggage that can shake up how employees live and feel outside of work.
The study revealed that working with A.I. creates attachment anxiety, which in turn increases people’s need for “affiliation,” or person-to-person connection. When workplace human interactions are replaced with computerized ones, people react both positively and negatively.
The positive outcome of people’s A.I.-induced craving for connection is that they become motivated to spend extra time with their colleagues and to voluntarily help each other. The downside is that outside of work, people feel more isolated, which can lead to serious consequences including a lack of sleep and more alcohol consumption.
Pok Man Tang, the study’s lead researcher and a professor at the University of Georgia business school who focuses on human-nonhuman interactions at work, told Fortune that he developed the research questions behind the study during his former career as a bank analyst at BNP Paribas.
“That’s the time when I observed a lot of interesting interactions and dynamics and behavioral consequences among my colleagues, that made me start wondering, ‘Why is it leading to this series of outcomes?’” Tang said.
Tang questioned why his colleagues drank more at happy hour when they used more A.I., and later discovered that the cause was A.I.-induced loneliness. The study said that consuming alcohol helps dull feelings of isolation and provides an escape from people’s awareness of loneliness. Specifically, after-work drinking is a “spillover” negative effect of lack of workplace connection.
Still, Tang emphasizes that A.I.’s impact on work isn’t purely harmful—rather, it’s best characterized by tension. It can make people feel at once more collegial and more detached from their coworkers, and it also has a complicated effect on workplace self-esteem. Using the technology isn’t the same as using any other tool, like a calculator or spell-check. For some, including Tang himself, interacting with A.I. is emotional and confusing.
“I had issues with A.I., personally. I experienced a lot of tensions intrapersonally while I was using it. I just struggled every moment,” Tang told Fortune. “It’s probably because paradoxical feelings are happening at the same time.”
The study likens the current moment in work to the 2009 Bruce Willis film Surrogates, in which humans are completely homebound while A.I. androids complete all work and outside tasks. Humans operate their A.I. “surrogates,” which are far more capable than they are, from the safety and comfort of their houses.
“This movie depicts a sensationalized example of which is becoming a commonplace workplace phenomenon—that many work interactions are not with human coworkers but rather with AI systems that give advice, make decisions, and increasingly are employees’ primary collaborators in the pursuit of work goals,” the study says.
Tang’s study was based on a survey of 800 workers in various fields across the U.S., Indonesia, Taiwan, and Malaysia. The findings were published in the Journal of Applied Psychology.
In previous research, Tang has found other examples of the tension A.I. engenders. In a 2022 report, he concluded that using A.I. makes workers feel more confidence in achieving their goals, but also more ambiguity toward what’s expected of them.
The technology makes tasks easier, or fully automates some tasks, which can cause psychological conflict. Employees’ workflow benefits, but they may question their value when technology assumes many of their responsibilities.
However, Tang speculates that the negative social consequences of A.I. may wane in coming decades. Currently, the technology’s integration is novel and therefore disruptive, but as A.I. becomes increasingly pervasive, people may be less jarred by its impact.
Companies may be able to incorporate A.I. in ways that support employee wellness rather than erode it. Tang said that management style is key to a smooth A.I. transition. Tying A.I. into a shared company identity and communicating actively with employees mitigates negative consequences, he added.
“I think some leaders tried to sell A.I. to their employees as ‘It’s a tool, it’s a machine, go and use it.’ I think that’s not a very constructive way of incorporating it,” Tang said. “A better way to do that is through different kinds of intervention, training, or communication from senior leaders. They have to convey that A.I. is actually part of the company’s shared vision and goal.”
