In summer 2024, software company Lattice announced some new hires of sorts: a cadre of AI “employees” the firm would onboard, train, and manage like human workers. Though the tech unicorn founded by Sam Altman’s brother ultimately walked back some of the “rights” for its digital employees following pushback after it laid off 15% of its human staff, the trend of AI agents popping up on organization charts has not dissipated. In fact, new research shows this practice has only gotten more popular—and it’s making human employees worse at their jobs as a result.
A study conducted by the Boston Consulting Group (BCG) found nearly one-third of managers across the U.S., Canada, and European Union framed AI as a teammate or employee, and more than 20% listed those AI agents on their company’s work charts.
But the study warned of the dangers of personifying these AI tools and treating them as one would a human employees. Researchers led by Matthew Kropp, a managing director and senior partner at BCG surveyed more than 1,200 human resources and finance professionals on how AI was used in the workplace and then asked them to assess a workplace document with multiple errors in it. The participants were given the same document, but assigned into three groups: one where the document was attributed to a human employee, one to an AI tool, and another to a named AI “employee.” Those in the group with the document attributed to the AI employee were able to identify fewer errors. They also reported less accountability, blaming the AI agent, rather than themselves, for a mistake, and also were more likely to ask another employee to review the work of the AI employee, making a colleague’s job harder.
“If you translate that into an organizational context, that means people sort of passing the buck,” Kropp told Fortune. “That’s creating more work for somebody else in the organization, and therefore you’re creating churn and excess overhead in the organization.”
AI has yet to show consensus productivity gains, and with companies not yet able to fundamentally transform business operations with the technology, skepticism over the technology’s promises to deliver its return on a projected $2.5 trillion in investments has grown
Previous research has already demonstrated the consequences of deploying AI without appropriate expectations or guardrails. One study published last year uncovered that experienced software engineers who used AI took longer to complete tasks than colleagues who did not as a result of extra time spent debugging AI-generated code. BCG previously found that when software engineers used too many AI tools, they experienced “AI brain fry,” feeling a sense of overwhelm and brain fog associated with more mistakes and a subsequent drain on productivity.
AI anthropomorphosis
According to Kropp, employees working with anthropomorphized AI tools were essentially becoming lazier because they felt they were able to shift accountability from themselves to the technology.
“AI doesn’t have responsibility,” Kropp said. “It isn’t a person. It can’t be hired or fired. There’s no performance reviews for AI. It’s just a piece of software. So there can’t actually be accountability for an AI. Some human person has to be responsible for that process, and so we can’t let people sort of transfer the responsibility mentally.”
Beyond becoming the office scapegoat, AI employees were actually counterproductive in many companies’ mission to increase AI adoption in the workplace. By assigning human traits to AI agents or integrating them into org charts, managers hoped to increase employee acceptance of these new technologies, Kropp said. Instead, the study found that participants in the group assigned an AI employee did not report higher intention to adopt AI. Rather, they reported a 7% higher concern AI would replace their roles and 10% lower trust level in how AI would be deployed in the workplace.
To be sure, Kropp said his research wasn’t an indictment of AI in the workplace at all, but rather guidance on how to integrate budding technology without sacrificing productivity and professionalism. He drew on solutions for AI brain fry, which didn’t include eradicating AI from the workplace, but rather encouraging employees to group tasks with AI into one part of the work day, or taking intentional breaks from work altogether.
Similarly, Kropp argued, AI agents can be leveraged, but their roles need to be clearly defined to ensure human employees don’t feel threatened, but also empowered to take accountability over their own mistakes, even if it includes errors related to how they use AI.
“We’re not saying you should not put AI on your org charter, you should not name your AI systems,” he said. “What we’re saying is, if you do that, you should be aware of the behavior change that creates for the rest of the organization, and you should be actively managed for that.”
