Three weeks ago I wrote This Is Your Brain on AI. I pulled together every study I could find on AI-driven cognitive fatigue — MIT’s EEG research, the METR developer study, Upwork’s burnout survey, UC Berkeley’s “workload creep” findings. I described the exhaustion I feel at the end of a day where AI has touched nearly everything — the buzzing, the mental fog, the sense that my brain had been processing information at a pace human biology wasn’t designed for.
But I also admitted something: the research didn’t quite match the experience. The MIT study measured people writing essays with ChatGPT. The METR study tracked whether developers were actually faster. The UC Berkeley work was observational. None of them captured the specific kind of drain that comes from doing twice as much as you used to — not because anyone asked you to, but because the tools made twice as much feel possible.
My day job is meetings and project management. I’m an IT director in healthcare. I don’t spend my 8-to-5 writing code — I jump at the chance when I can, but the reality is back-to-back meetings, vendor calls, and project coordination. AI hasn’t changed that calendar. What it’s changed is what I can do with the margins. Every meeting gets documented. Action items get tracked in a structured GitHub repo. Follow-up emails get drafted in seconds. Every meeting gets documented. Action items roll forward from one meeting to the next, so nothing mentioned once disappears into the ether. I even built a tool to help capture meeting conversations in real time. I can take in, consolidate, and share more information than I ever could before.
But I can’t make more phone calls. I can’t sit in more meetings than the calendar allows. I can’t make a vendor respond faster or a project dependency resolve itself sooner. The human-paced parts of the job haven’t sped up at all — and yet there’s this nagging sense that they should have, just because everything around them did.
Now there’s a study that names what that feels like.
The BCG Study
Researchers from Boston Consulting Group and UC Riverside surveyed nearly 1,500 full-time U.S. workers and published their findings in Harvard Business Review this month. They coined the term “AI brain fry” and defined it as something specific: mental fatigue from excessive interaction with and oversight of AI tools beyond one’s cognitive capacity.
The key move: they distinguished brain fry from burnout. Burnout is emotional exhaustion — the slow erosion of caring about your work. Brain fry is acute cognitive overload — your brain hitting a wall because it’s processing too much, too fast, from too many directions. They’re related but they’re not the same thing. You can be deeply engaged and cognitively fried at the same time.
That distinction is exactly what I was trying to articulate in the previous post. The exhaustion I described isn’t apathy. It’s the opposite — it comes from being too engaged, moving too fast, reviewing too much output.
The Numbers
Fourteen percent of AI-using workers reported brain fry. That ranged from 6% in legal to 26% in marketing. Software development, HR, finance, and IT were all in the upper range.
The findings that jumped out at me:
Productivity peaks at two AI tools, then declines. One tool helped. Two showed significant gains. Three showed diminishing returns. Beyond three, productivity scores actually went down. That’s a concrete threshold nobody had quantified before — and it maps to what I’ve observed. There’s a point where adding another agent or another tool tips you from “force multiplier” into “air traffic controller with too many planes.”
Major errors increased 39%. Not minor typos — major errors. The kind that matter. This connects the subjective feeling of being overwhelmed to objective quality degradation. You’re not imagining it. When you’re cognitively fried, your work gets worse.
Decision fatigue increased 33%. Every AI output requires a decision: accept, reject, modify, investigate further. When you’re already in eight meetings a day and using AI to keep up with the throughput between them, you’re burning through decision-making capacity at a rate that didn’t exist two years ago.
Intent to quit rose 39% among the most affected. Brain fry isn’t just uncomfortable — it’s a retention problem. The people most likely to experience it are high performers, the ones organizations can least afford to lose.
What Managers Get Wrong
The study found two management behaviors that made things measurably worse. First, expecting employees to figure out AI tools on their own — self-teaching predicted 5% higher mental fatigue across the team. Second, telling employees that AI should increase their workload — that predicted 12% higher fatigue.
The intervention that worked? Managers who simply made themselves available to answer AI-related questions saw 15% lower fatigue on their teams. Not training programs, not mandates. Just availability.
The Reddit Thread Says the Quiet Part
A Reddit thread discussing the study drew responses from people living this daily. The developers focused on code review fatigue, but one comment captured something more universal:
Your brain has this concept of what a day’s work should be, and when that’s met by 10am, your brain goes into shutdown mode — but you still have seven hours left.
That resonates beyond coding. When AI helps you clear your inbox, document three meetings, draft a project update, and triage a vendor issue before lunch, you’ve done what used to be a full day. But the calendar doesn’t care. The afternoon meetings still happen. The decisions still need making. And your brain is already cooked.
Another commenter put it simply: it “feels less like ‘AI made work easier’ and more like it moved the cognitive load from production to supervision.” That’s the shift. I’m not doing less thinking. I’m doing a different, denser kind of thinking — reviewing, validating, deciding — on top of everything else.
What I’m Doing About It
I don’t have a grand solution. I’m not pacing myself. I’m cramming as much activity into every day as I can, hoping it’s enough to meet expectations in an AI-accelerated world. I’ve done more in the last year than in any single year of my 30 years in IT — and I’m not sure it’s sustainable, but I’m not sure slowing down is an option either.
Just because I can document every meeting, maintain a structured repo for all of it, draft every follow-up, and still have time to write a blog post in the evening doesn’t mean my brain isn’t paying a price for the throughput.
The research finally caught up. The egg is in the pan. Now we just need to stop pretending we don’t smell it burning.