The first real signs of AI burnout are not coming from workers afraid of artificial intelligence, but from those who have fully embraced it.

For years, the dominant message in American work culture has been reassuring: AI won’t replace most people — it will save them. While some white-collar roles may disappear, the broader promise has been that AI acts as a force multiplier, making professionals more productive and indispensable while reducing effort and workload.

A new study published in Harvard Business Review challenges that narrative. Instead of a productivity revolution, researchers warn that companies may be turning into burnout engines.

When Productivity Turns Into Overwork

As part of what they describe as “in-progress research,” UC Berkeley researchers spent eight months inside a 200-person tech company observing what happened when employees genuinely adopted AI tools.

Across more than 40 in-depth interviews, researchers found no pressure from management to increase output or meet higher targets. Instead, workers voluntarily took on more tasks simply because AI made more work feel possible.

That shift fueled AI burnout. Work spilled into lunch breaks, evenings, and personal time. As AI freed up hours, those hours were quickly filled — and then exceeded.

One engineer summarized the experience bluntly: “You don’t work less. You just work the same amount or even more.”

Rising Expectations Without Real Relief

Similar sentiments have surfaced elsewhere. On the tech forum Hacker News, one commenter described how expectations soared after their team adopted AI tools.

“Expectations have tripled, stress has tripled, and actual productivity has only gone up by maybe 10%,” the commenter wrote, adding that pressure to justify leadership’s AI investments led to longer working hours.

The pattern shows how AI burnout can take hold even without explicit demands, driven instead by cultural expectations for constant responsiveness and speed.

What Happens When AI Actually Works

Previous research has questioned whether AI delivers meaningful productivity gains. Some studies found developers took longer to complete tasks despite believing they were faster, while broader workplace research showed only modest time savings with no impact on hours worked.

This study takes a different approach. It accepts that AI can augment human capability — then examines the consequences.

According to the researchers, that augmentation often leads to “fatigue, burnout, and a growing sense that work is harder to step away from,” especially as expectations around availability and speed increase.

A New Risk in the AI Era

The industry wagered that helping people do more would solve workplace challenges. The early evidence suggests it may instead be creating a new one.

As AI burnout spreads among the most enthusiastic adopters, the promise of productivity without pressure is being tested — and the results point to a growing problem that may define the next phase of AI at work.