A survey of 3,000 daily AI users finds the technology is delivering real gains while generating a form of cognitive fatigue few organizations are measuring. Researchers have a name for it: “AI brain fry.”
June 2026
For most of the past two years, the defining question in artificial intelligence has been one of capability: can the model write the email, summarize the report, pass the benchmark. Increasingly, it can. A new study suggests the more important question for the years ahead is a different one — what sustained, daily use of these tools does to the people relying on them.
The research, from AI platform Getsolved, surveyed 3,000 Gen Z professionals aged 18 to 29 who use AI every day. It documents a pattern the researchers call “AI brain fry”: cognitive fatigue that stems not from difficult work, but from the constant effort of monitoring, verifying, and correcting what AI produces.
The productivity case is not in doubt. Three in four respondents said AI had made them more productive, and most reported improvement in problem-solving (65%), creativity (62%), focus (61%), and critical thinking (54%). One skill stood apart: memory, which improved for fewer than half — the lowest of any measured. When a tool summarizes, stores, and retrieves information on your behalf, the researchers note, the incentive to retain it yourself quietly disappears.
That outlier is the one Harry Southworth, Head of AI Development at Getsolved, keeps returning to. “We’ve effectively solved the capability question — these systems can do an enormous amount,” he said. “The open question now is sustainability: what happens to people who work this way every day, for years. Memory being the one skill that didn’t improve is an early hint that offloading cognition has trade-offs we don’t fully understand yet.”
The fatigue data fills in the rest of the picture. More than a third of respondents reported near-daily mental fog, a similar share felt drained even when the work was easy, and nearly one in four said AI had hurt their mental health. Recovery is uneven — 41% need a full evening of rest after an AI-heavy day, and roughly one in ten carries unresolved tiredness into the next. Yet 87% described themselves as neutral or energized, a normalization the researchers warn makes the problem easy to miss.
For Southworth, the finding points to a gap in how the industry thinks about responsibility. “Responsible-AI conversations have focused on bias, safety, accuracy — all essential,” he said. “But there’s a human-factors dimension we’ve barely started on. If a tool makes someone 20% faster and meaningfully more drained, a mature industry measures both sides of that ledger, not just the side that looks good in a demo.”
Organizations, for now, are mostly measuring one side. Just over half of respondents said their employer does not address the cognitive load of AI work, and only 39% said their company actively tries to manage it. With 41% reporting anxiety about AI at work, the distance between how fast AI is being deployed and how well its effects are understood appears to be widening.
None of this argues against AI adoption — the people in the study are among its heaviest and most capable users. It argues for a more complete definition of what “working” means. The AI platform’s full findings suggest the field’s next milestone won’t be a more capable model, but a more sustainable way of working alongside one. For an industry that has spent years proving what AI can do, the harder and more valuable work may be figuring out what it should ask of the humans in the loop.
