A new study by researchers at Duke University has revealed that employees who use AI tools such as ChatGPT, Gemini or Copilot at work are often perceived as less intelligent, less diligent, and even lazier than those who don’t. While artificial intelligence is increasingly credited with enhancing productivity, this research highlights a potential social cost that could slow its mainstream adoption in workplaces.
Published in the Proceedings of the National Academy of Sciences, the study was conducted by Jessica Reif, Richard Larrick and Jack Soll. It involved four online experiments with 4,400 participants, designed to examine how people view colleagues who use AI applications for work-related tasks.
In the first experiment, participants were asked to imagine themselves using an AI app to complete a project, and then predict how co-workers would perceive them. The majority believed they would be judged as lazy, incompetent, or easily replaceable.
The second experiment flipped the perspective, asking respondents to describe colleagues who used AI tools to get work done. Again, the findings were consistent: AI users were seen as less competent, less self-assured, and lacking independence.
A third scenario positioned participants as hiring managers. They were asked to evaluate job candidates, some of whom admitted to using AI to complete their work. Candidates who disclosed AI use were generally rated less favourably unless the hiring managers themselves had experience using AI, in which case the bias largely disappeared.
In the final experiment, researchers tested perceptions when the AI tool was known to be appropriate for the task and demonstrably improved productivity. Under these circumstances, the negative judgments were significantly reduced.
Interestingly, the study found a consistent pattern: those with personal experience using AI were much more accepting of it in themselves and others.
Taken together, the findings suggest that while AI tools may offer clear functional benefits, their adoption in the workplace could be hampered by lingering social biases. For AI to be fully embraced in professional settings, workplace culture may need to evolve alongside the technology.