Google Research and DeepMind study reveals challenges of AI agents in workplace

A five-month study by Google Research and Google DeepMind on the Team Agent system across over 20 teams revealed significant friction between AI assistants and human employees. The AI agents struggled with office hierarchy, such as assigning tasks to executives, and failed to distinguish between current and outdated project documentation. These issues led to productivity losses, with some employees opting to move conversations to private channels to avoid the bot's disruptive notifications.