If you are anything like the managers recently studied by Emma Wiles, a Boston University business professor, treating an AI agent as a “coworker” rather than a software tool could actually make you do a worse job. Wiles found that people caught 18% fewer errors when the work was described as coming from an agentic “AI employee” instead of a chatbot. It turns out that what you call something really does matter.
This is an alarming glimpse into the future that Silicon Valley is rushing us toward. Last year, Nvidia’s CEO Jensen Huang talked about workplaces filled with “digital humans.” Since April, Microsoft, OpenAI, Anthropic and Google have all released new tools oriented toward managing teams of AI agents — several of which are explicitly marketed as digital colleagues with the flexibility and cognitive power of real humans. And nearly a third of the 1,261 managers who participated in Wiles’s study reported that their companies already frame AI agents as employees (23% even list them on org charts).
The technical progress of agentic AI is not all hot air, of course. Agents — which are essentially AI tools programmed to work in a loop until they accomplish a goal — have become measurably better at more complex tasks. However, it is a huge leap to refer to these tools as coworkers or employees, and doing so sets unrealistic expectations for what AI can do while leaving the human workers supposedly overseeing them worse off.
That is partly because, as Wiles’s research suggests, it reverses our sense of responsibility. When an AI tool was framed as an employee, participants in the study viewed themselves as less responsible for its output. They were also 44% more likely to escalate its questionable work to a manager for further review instead of trusting their own corrections — thus defeating the time-saving purpose of using the AI agent in the first place.
This matters far beyond workplace culture. As AI agents are embedded into healthcare, military operations, education and government, there is a growing risk they will become a convenient place to offload blame for failures that are actually the result of poor human decisions, incentives and oversight. (Remember how the bomb strike on a women’s school in Iran was widely blamed on Claude, when all evidence points to a cascade of human errors.)
“AI agents right now are being marketed as things that can replace humans, and I think that’s just a losing proposition,” says Daron Acemoglu, an MIT economist who won the Nobel Prize in 2024 and studies AI’s impact on the economy. “They should instead be optimized so they’ll augment human capabilities, which is not what they’ve been doing currently.”
What would that look like? Consider a recent effort at Stanford, where researchers gave 1,500 workers across 104 jobs information about what tasks AI could potentially handle in their work and then asked what would actually be most useful and productive. Workers did want automation in certain areas: law clerks thought AI could help ensure adequate progress was being made across cases, for example. But overall, the tasks that tech consultants deemed most suitable for AI — such as verifying customer credit scores for sales reps — were exactly the things the actual workers said they did not want or need an agent to do.
Which brings us back to Alex. Calling Alex an employee is easy — and convenient, especially when something goes wrong — but it is a branding exercise. It does not make the tool any better at the job, and as Wiles’s research shows, it makes the humans around it worse at theirs. And remember: they are the ones with the agency that AI is trying to replicate. They deserve better than Alex.
