MIT Technology Review is reporting that new research suggests treating artificial intelligence tools as "coworkers" or "employees" can lead to worse human performance. It said a study by Boston University business professor Emma Wiles found that managers caught 18% fewer errors when work was attributed to an "AI employee" named Alex compared to a chatbot. The outlet noted that this framing also made participants 44% more likely to escalate questionable AI work to a manager rather than correcting it themselves, inverting the sense of responsibility.
The publication highlighted a trend in Silicon Valley, with companies like Nvidia, Microsoft, OpenAI, Anthropic and Google releasing tools marketed as "digital humans" or AI agents with human-like flexibility. It said nearly a third of the 1,261 managers in Wiles's study indicated their companies already frame AI agents as employees, with 23% even listing them on organizational charts.
MIT Technology Review said this approach sets unrealistic expectations for AI and can leave human employees worse off, potentially leading to misattribution of blame for failures in critical sectors like health care and warfare. Daron Acemoglu, a Nobel Prize-winning economist at MIT, said AI agents should be optimized to improve human capabilities rather than replace them. The outlet also mentioned Stanford research indicating that workers often desire different AI assistance than what tech experts deem suitable.
Full Article: AI agents are not your “coworkers”