New research finds that while three in four young professionals say AI boosts their productivity, more than half have stopped using it at times because supervising the technology is too mentally exhausting.
June 2026
Every productivity pitch for artificial intelligence sounds roughly the same: do more, faster, with less effort. For the most part, the people using it agree. But a new body of research suggests that the same tools delivering those gains are also producing a quieter, less convenient effect — cognitive fatigue that most companies, and much of the tech industry building these tools, have yet to take seriously.
The data comes from a survey by Getsolved, an AI platform whose tools help people check and refine written work. The company polled 3,000 professionals aged 18 to 29 who use AI daily, and its findings describe a striking gap between how productive AI makes people feel and what it quietly costs them.
On the surface, the results read as a strong endorsement. Three in four respondents said AI had made them more productive, and 60% said it helped them work faster with less effort. Most also reported gains in problem-solving, creativity, and focus.
The complication appears when workers are asked why they sometimes don’t use it. More than half — 52% — said they had avoided AI specifically because supervising and correcting it felt too mentally draining. One in four said AI producing wrong information had actively made their jobs harder, and others pointed to the constant need to fact-check as the real friction: not that AI is occasionally wrong, but that it could be wrong at any moment, so every output demands review.
For Harry Southworth, Head of AI Development at Getsolved, that verification burden is the part the industry has overlooked. “We’ve spent two years optimizing for output, and it’s worked,” he said. “What we haven’t optimized for is the person on the other side who has to check, correct, and trust what the system produces. That cost doesn’t show up in a benchmark, but it shows up in whether people keep using the tool.”
The strain isn’t only about wasted time. In the same survey, 36% said they deal with mental fog or trouble concentrating almost every day, 37% feel worn out even when the work itself wasn’t demanding, and nearly one in four said AI at work has negatively affected their mental health.
After an AI-heavy day, 41% said they need a full evening of rest to recover.
Curiously, most workers didn’t describe themselves as struggling. Even as they reported those symptoms, 87% said they felt neutral or energized after AI-intensive days — a disconnect Southworth finds telling. “The most striking thing was that people felt fine even while reporting daily fog and fatigue,” he said. “When strain becomes the baseline, it stops being a signal anyone acts on. For anyone building these tools, that should be a wake-up call: your users may be struggling in ways they’ll never report.”
Few workplaces are positioned to catch it. Just over half of respondents said their employer does not address the mental load of AI work at all, and only 39% said their company genuinely tries to manage it. About one in ten said their employer was pushing for more AI use without acknowledging any cost.
The takeaway for the companies building AI isn’t that the productivity gains are fake; the research is clear that they’re real. It’s that the next competitive frontier may not be raw capability at all, but trust — tools that are easier to verify, easier to rely on, and less taxing to oversee. As the full findings of the AI platform suggest, the workers most fluent with AI are already voting with their attention. The question is whether the industry is listening.
