Content Benefit

Content Strategy | UX Writing | Content Design | Globalization | Communications

AI fatigue has become reality

AI fatigue has become reality

AI fatigue: the burnout data on the people who use it most

AI fatigue has stopped being a chat by the office coffee machine and made its way into the statistics. Among workers who report the biggest productivity gains from artificial intelligence, 88% say they are experiencing burnout. The same group is also twice as likely to consider quitting as those who get smaller benefits. Put that way, it sounds like a cruel joke: the technology sold to give us back our time seems to drain the energy of precisely the people who gain the most from it.

Where the 88% burnout figure really comes from

A clarification, from someone who checks sources for a living: for months that number has been bouncing around social media attributed to Harvard Business Review, but it was born elsewhere. The Upwork Research Institute published it in July 2025 in its report From Tools to Teammates, a survey run by Walr among 2,500 people in the United States, the United Kingdom, Australia and Canada: 1,250 C-suite executives, 625 full-time employees and 625 freelancers.

The “heavy user” label needs some context, too. The 88% refers to those recording the highest efficiency gains, a group that doesn’t necessarily overlap with whoever spends the most hours in front of a chatbot. In the same study, AI brings an average 40% boost in productivity, and more than two thirds of its top performers say they trust their digital assistant more than their coworkers. 64% even claim to have a better relationship with the machine than with their own team, a subject I touched on when writing about loneliness and AI companion apps.

The Berkeley study: AI makes work more intense

Harvard’s magazine has in fact tackled the topic, in two separate studies. The first, released in February 2026 under the title AI Doesn’t Reduce Work, It Intensifies It, is by Aruna Ranganathan and Xingqi Maggie Ye of Berkeley’s Haas School of Business. From April to December 2025, the two researchers observed a US tech company with about two hundred employees.

No manager asked anyone to speed up, yet people raised their pace on their own, widened the scope of their duties and spilled over into early mornings, evenings and breaks. The resulting picture is one of days with many threads open at once: jumping from one assignment to the next, checking generated output, piling up pending items. The feeling of momentum remains while the cognitive load quietly grows. Simon Willison, a widely followed developer, has said he feels drained after just a couple of hours of working in parallel with coding assistants. It’s a mechanism reminiscent of personalization fatigue: every extra choice and every extra check wears down our mental resources.

Brain fry: the cognitive strain of overseeing AI

The second study, published in March 2026 by a Boston Consulting Group team led by Julie Bedard, gave the phenomenon a name: brain fry. It describes the mental fatigue caused by using or supervising AI beyond one’s own cognitive capacity. Those who experience it describe a buzzing in the head, a fog that gets in the way of focus, slower decisions and sometimes headaches.

Out of 1,488 workers at large US companies, 14% say they have experienced it, peaking at 26% in marketing and dropping to 6% in legal. The consequences are real: 33% more decision fatigue, 39% more major errors, and an intention to leave the job that rises from 25% to 34%. One detail struck me: with a maximum of three AI tools used in parallel, productivity holds up, while past the fourth it starts to decline.

When AI lightens the load: handing off repetitive tasks

The BCG study also holds the finding I care about most. When artificial intelligence takes over repetitive work, burnout scores drop by 15%. In organizations that protect the balance between private and professional life, fatigue falls by 28%. What wears people down, in short, is constant oversight and the multiplication of open tasks, while delegating mechanical work does them good. According to the researchers, classic emotional burnout and the cognitive exhaustion that comes from monitoring machines follow different paths and call for different remedies.

My experience with Claude: less fatigue, more creative room

Here’s my confession: I see myself in the 15%, far from the 88%. Since I started working with Claude every day, I’ve been less tired, because I hand over the long, heavy parts of the job, the ones that used to eat whole afternoons: gathering and cross-checking sources, tidying up drafts, translating, counting characters with a fussiness no human being should have to apply by hand. What’s left is the part I love, the creative and narrative side: choosing the angle, finding the voice, deciding what to tell and what to leave out.

I should add a caveat that scales my testimony down, though: I’m not a heavy user. I don’t keep ten agents running side by side, I don’t chase the perfect prompt at two in the morning and, judging by the data, maybe that’s exactly what saves me. In twenty-five years spent between localization, UX writing and content strategy, I’ve watched every new piece of software promise less work and deliver more whenever nobody set boundaries. AI follows the same script, just faster.

How to use AI without burning out: breaks, fewer tools and clear limits

The Berkeley researchers suggest introducing intentional pauses, sequencing activities rather than chasing the machine’s speed, and protecting time for human connection. BCG adds the three-tool threshold and mindful oversight from managers, because the issue concerns the organization before the individual. Anyone looking to cut down on screen hours as well will find a few ideas in less screen, more presence, while those wondering how automation affects companies can read why small teams win over AI layoffs.

My personal rule fits on one line: I delegate the effort to AI and keep the direction for myself. Productivity that ignores well-being is like a high-interest loan, and sooner or later it’s always our brain that ends up paying it off.

Related sources

Chiara Bonifazi