AI agents are not your "coworkers" (www.technologyreview.com)

A recent study by Emma Wiles, a Boston University business professor, reveals that framing AI tools as "coworkers" leads to poorer performance among human managers. Participants caught 18% fewer errors when interacting with an AI labeled as a "digital employee," demonstrating that terminology significantly influences expectations and responsibility. Nearly a third of the 1,261 managers surveyed reported that their companies categorize AI agents as employees, which could escalate accountability issues, especially in critical sectors like healthcare and education. The study warns that referring to AI as "coworkers" sets unrealistic expectations and undermines human oversight, as individuals become less responsible for AI-generated outputs. As tech giants like Microsoft and Google promote AI agents with human-like capabilities, experts like MIT economist Daron Acemoglu advocate for a shift in focus towards optimizing AI to enhance human capabilities rather than replace them. This perspective emphasizes the need for a redefined relationship between humans and AI, stressing the importance of maintaining human agency and responsibility in the evolving technological landscape.