Managers are letting AI coast in the office

Niv Bavarsky

On Tuesdays, the Brew’s Matty Merritt brings you the news you need to make life a little easier during your 9-5, 5-9, or OOO. Your AI “coworker” might be getting more latitude than you. A new study from Boston University professor Emma Wiles found that managers were less thorough in reviewing work from AI “employees”: Researchers gave error-filled memos to 1,261 managers, telling different groups that the documents were produced by AI tools, AI “employees,” or human employees.Participants who were already managing anthropomorphized agents at their jobs caught 18% fewer errors when they were proofing what they thought was the work of AI “employees.” Not my problem. Wiles chalked it up to a lack of understanding around who is responsible for errors made by chatbots.