The first signs of burnout come from those who embrace AI the most.

The most fascinating story in American work culture right now isn’t that AI will take your job. AI will save you from that.

That’s the version the industry has been selling for the past three years to millions of nervous people who want to buy it. Yes, some white-collar jobs will disappear. However, there is an argument to be made that in most other roles, AI is a force multiplier. You become a more capable and more essential lawyer, consultant, writer, coder, financial analyst, etc. The tools work for you, the less work you do, and everyone wins.

But a new study published in Harvard Business Review builds on this premise to draw a real conclusion, finding that there is no productivity revolution. This finds businesses at risk of becoming burnout machines.

As part of what they describe as “an ongoing study,” researchers at UC Berkeley spent eight months at a 200-employee technology company observing what happened when employees truly embraced AI. What they found, through over 40 ‘in-depth’ interviews, was that no one in this company was under pressure. No one was told to achieve new goals. People started doing more because tools allowed them to do more. But because I was able to do this, I started working through lunch and late into the evening. Employees’ to-do lists expanded to fill each hour AI freed up, and the process continued.

One engineer told them, “You might have thought that increasing your productivity with AI would save you time and do less work. But in reality, you don’t do less work. You just have to do the same amount of work, or more.”

“It feels like this: Since our team shifted all of our work styles to AI, expectations have tripled, stress has tripled, and actual productivity has only increased by about 10%. It feels like leadership is putting enormous pressure on everyone to prove that the investment in AI is worth it, and we all feel pressured to try to show that it’s true by actually working longer hours to make it happen.”

It’s both fascinating and surprising. The debate about AI and work has always stalled on the same question: Are there really any benefits? But too few people stop to ask what’s happening.

The researchers’ new findings aren’t entirely new. A separate trial last summer found that skilled developers using AI tools took 19% longer to complete tasks, even though they believed they were 20% faster. A National Bureau of Economic Research study that tracked AI adoption in thousands of workplaces around the same time found that productivity gains were only 3% in terms of time savings and had no significant impact on earnings or hours worked in any occupation. Both studies were separate.

This may be harder to ignore because it doesn’t challenge the premise that AI can augment what employees can do on their own. According to the researchers, they confirm this and then show what the augmentation actually brings. This is “fatigue, burnout, and an increasing perception that it is more difficult to disengage from work, especially as organizational expectations for speed and responsiveness increase.”

The industry is convinced that helping people do more will be the answer to everything, but it could also be the start of a whole different problem. The study is worth reading here.