A Harvard Business Review study by Berkeley Haas researchers reveals AI doesn't reduce workloads but intensifies them. The 2025 study of 200 tech employees found workers managing multiple active threads simultaneously, manually writing code while AI generates alternatives, and reviving long-deferred tasks. This creates a false sense of productivity while increasing cognitive load.

Workers described AI as a "partner" that helps move through workloads, but the reality involves constant attention switching and frequent AI output verification. Researchers documented employees juggling numerous open tasks simultaneously, leading to mental exhaustion despite completing more work. The study captured a phenomenon observed by many developers using LLMs in professional settings.

The HBR piece calls for organizations to develop structured AI practices to prevent burnout and distinguish genuine productivity gains from unsustainable intensity. Workers report losing sleep as they find adding "just one more prompt" irresistible, disrupting decades of established working practices. Finding a sustainable balance with AI tools requires discipline and organizational intervention.