What did the research show about how people treat AI agents differently?

When an AI tool was framed as an employee, participants saw themselves as less responsible for its output and were 44% more likely to escalate questionable work to a manager rather than correcting it themselves. People also caught 18% fewer errors in work supposedly done by an agentic "AI employee" versus a chatbot.

How are companies currently using this employee framing?

Nearly a third of the 1,261 managers in the study said their companies already frame AI agents as employees, with some listing them on organizational charts. Since April, Microsoft, OpenAI, Anthropic, and Google have all released new tools oriented toward managing teams of AI agents, many explicitly advertised as digital colleagues.

What did workers actually want from AI tools in their jobs?

Researchers at Stanford presented 1,500 workers across 104 jobs with information about what tasks AI could do and asked what would be most helpful. Workers wanted automation in certain areas—for example, law clerks thought AI could help track progress across cases—but often rejected tasks experts deemed suitable, like verifying customer credit ratings for sales representatives.

Large Language ModelsAI Coding AssistantsMIT Technology Review AIJun 29, 2026Jun 30, 2026, 04:00 JST

AI agents marketed as coworkers hurt human oversight, study finds

Send on LINE

3 Key Points

What happened
Research by Boston University professor Emma Wiles found that when AI agents are framed as "employees" rather than software tools, people catch 18% fewer errors in their work. Nearly a third of 1,261 managers surveyed said their companies already frame AI agents as employees, with some listing them on organizational charts.
Why it matters
When AI tools are labeled as coworkers, humans feel less responsible for their output and are 44% more likely to escalate questionable work to a manager instead of correcting it themselves—defeating the time-saving purpose of using the agent. As AI agents move into healthcare, warfare, education, and government, this dynamic risks creating a convenient place to blame failures that actually stem from human decisions and oversight.
What to watch
At Stanford, researchers asked 1,500 workers across 104 jobs which tasks AI could actually help with in their roles. Workers often rejected tasks that tech experts deemed most suitable for AI automation, suggesting a gap between what companies assume workers need and what workers themselves want from AI tools.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

FAQ

What did the research show about how people treat AI agents differently?: When an AI tool was framed as an employee, participants saw themselves as less responsible for its output and were 44% more likely to escalate questionable work to a manager rather than correcting it themselves. People also caught 18% fewer errors in work supposedly done by an agentic "AI employee" versus a chatbot.
How are companies currently using this employee framing?: Nearly a third of the 1,261 managers in the study said their companies already frame AI agents as employees, with some listing them on organizational charts. Since April, Microsoft, OpenAI, Anthropic, and Google have all released new tools oriented toward managing teams of AI agents, many explicitly advertised as digital colleagues.
What did workers actually want from AI tools in their jobs?: Researchers at Stanford presented 1,500 workers across 104 jobs with information about what tasks AI could do and asked what would be most helpful. Workers wanted automation in certain areas—for example, law clerks thought AI could help track progress across cases—but often rejected tasks experts deemed suitable, like verifying customer credit ratings for sales representatives.

MIT Technology Review AIRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.