Artificial intelligence tools are increasingly promoted as a way to make work easier and reduce pressure on employees. However, new research published in Harvard Business Review suggests that productivity gains from AI may instead be expanding workloads and contributing to higher burnout risk across knowledge-based jobs.
The study, described as in-progress research, was conducted by researchers from the University of California, Berkeley. Over an eight-month period, the team observed how employees at a 200-person technology company integrated AI tools into daily work. The researchers conducted more than 40 in-depth interviews and closely tracked how expectations and work habits evolved as AI adoption increased.
More Capacity, More Work
According to the findings, employees were not explicitly directed by management to work longer hours or meet higher targets. Instead, the availability of AI tools made additional tasks feel achievable, leading workers to voluntarily take on more responsibilities. As a result, work increasingly extended into lunch breaks, evenings, and personal time.
One engineer interviewed by the researchers said that while AI initially appeared to save time, it did not ultimately reduce working hours. Instead, employees reported working the same amount or more, as new tasks and expectations quickly filled any time savings created by automation. The researchers concluded that AI-enabled efficiency often led to expanded to-do lists rather than meaningful reductions in workload.
Disconnect Between Perception and Output
The findings align with earlier research that has questioned the scale of AI-driven productivity gains. A trial conducted last summer found that experienced software developers using AI tools took 19% longer to complete tasks, despite believing they were 20% faster. Separately, a National Bureau of Economic Research study examining AI adoption across thousands of workplaces found average productivity gains of about 3% in time savings, with no significant impact on earnings or hours worked.
While those studies have faced methodological criticism, the Berkeley research differs in that it does not dispute AI’s ability to enhance individual capability. Instead, it suggests that increased capacity may create pressure to deliver more output, particularly as organizational expectations for speed and responsiveness rise.
Rising Expectations and Burnout Risk
The researchers found that as AI tools became embedded in daily workflows, employees felt a growing obligation to demonstrate the value of those tools. This dynamic contributed to stress and difficulty disengaging from work, especially in environments where rapid response times are implicitly rewarded.
According to the study, the outcome of this cycle is not necessarily a productivity breakthrough but a pattern of fatigue and burnout. The researchers warned that without deliberate limits, AI adoption could unintentionally turn workplaces into “burnout machines,” where efficiency gains lead to escalating demands rather than sustainable workloads.
Implications for Employers
The findings highlight a growing challenge for companies investing heavily in AI technologies. While automation can expand what individual workers are capable of doing, the research suggests organizations may need new management practices to prevent workload inflation. Without structural changes, AI’s promise of freeing up time may instead result in longer hours and higher stress, reshaping work in ways that were not originally intended.