What did the research reveal?
Generative artificial intelligence is often promoted as a tool that reduces workload. However, a study conducted by Berkeley Haas School of Business reached a different conclusion: integrating AI into daily operations frequently leads to a steady increase in labor intensity.
The research, led by Associate Professor Aruna Ranganathan and doctoral researcher Maggie Sinci, followed a U.S.-based technology company with around 200 employees over eight months. The team observed engineers, designers, product managers, and operations staff, analyzed internal communication channels, and conducted more than 40 in-depth interviews.
How did task scope expand?
The first form of labor intensification identified was task expansion. By filling knowledge gaps quickly, AI enabled employees to take on responsibilities that previously belonged to other specialists.
Product managers and designers began writing code, while researchers assumed engineering-related duties. As a result, professional boundaries widened and overall responsibility increased.
Why did work-life boundaries become blurred?
The second trend was the erosion of clear boundaries between work and personal time. Because AI tools are easy to access, employees often submitted prompts during lunch breaks, short pauses, or late in the evening.
These micro-interactions may not feel like full work sessions, yet collectively they increase workload and keep individuals mentally engaged in job-related tasks for longer periods.
How did multitasking intensify?
The third pattern involved rising multitasking demands. Running multiple AI agents simultaneously and constantly switching between different task streams created continuous cognitive pressure.
Frequent shifts in attention accelerate mental fatigue and may negatively affect the quality of decision-making over time.
What self-reinforcing cycle emerged?
As AI sped up task completion, expectations around speed increased. Higher expectations strengthened reliance on AI, which in turn broadened the range of tasks employees handled. This expansion further intensified work density.
One engineer described the paradox by noting that while AI was expected to save time and reduce effort, in practice employees ended up working the same amount — or even more.
What does short-term productivity growth conceal?
The researchers caution that immediate gains in productivity can mask a growing overload. Over time, this may lead to cognitive exhaustion, burnout, and a decline in the overall quality of decisions.
What solutions do researchers propose?
As a counterbalance, the authors recommend developing an “AI practice” — a structured set of norms and procedures governing how AI tools are used within organizations.
Suggested measures include mandatory pauses before major decisions, batching notifications into scheduled deliveries to reduce attention fragmentation, and allocating time for in-person collaboration. Such practices help restore shared understanding and support creative thinking.
How widespread is AI adoption?
Between the second and third quarters of 2025, the share of U.S. employees using artificial intelligence at work at least several times a year increased from 40% to 45%. This trend highlights the accelerating integration of AI into professional environments.