Why Generative AI Improves Productivity Without Shortening Working Hours

Researchers at the University of California, Berkeley’s Haas School of Business tracked a technology company with about 200 employees for eight months. They found that while generative AI accelerated document preparation, coding and analysis, employees used the time saved to take on more cross-functional tasks, with work spilling into after-hours time. The issue lies not only with the tools, but also with companies’ failure to redesign workloads, collaboration processes and performance standards accordingly. The study did not include any financial estimates.

Harvard Business Review published the Berkeley Haas research on February 9, 2026. The sample comprised 40 people observed from April through December 2025. On June 10, TechNews cited examples from Amazon and Google: documents that once took more than one hour were completed in 15–20 minutes, while summaries of six months of meetings went from taking 1–2 hours to 5–10 minutes. But the time saved was redirected to new tasks, leaving working hours unchanged.