Calling Alex a coworker made humans miss 18% more errors
a boston university study found managers caught 18% fewer errors when ai tools were framed as 'employees' rather than software, suggesting anthropomorphizing ai increases oversight risks and reduces accountability.
a study of 1,261 managers by boston university professor emma wiles found that framing ai tools as 'employees' led humans to catch 18% fewer errors compared to framing them as chatbots. participants were also 44% more likely to escalate work instead of correcting it themselves, indicating that anthropomorphizing ai reduces user accountability. this challenges the current trend where microsoft, google, and nvidia market 'digital humans' as colleagues. mit economist daron acemoglu warns that marketing agents as human replacements is a 'losing proposition' compared to tools designed to augment human capability.