AI Brain Fry: The Exhaustion Nobody Warned You About
AI Brain Fry: The Exhaustion Nobody Warned You About
The pitch goes like this: AI agents handle the grunt work, you think strategically, maybe leave early. It’s a nice idea.
Boston Consulting Group just published a study through Harvard Business Review that names what actually happens instead. They call it “AI brain fry” — mental fatigue “from excessive use or oversight of AI tools beyond one’s cognitive capacity” (CNN Business, March 13, 2026).
”A dozen browser tabs in my head”
Study participants who managed multiple AI agents reported a distinct cognitive sensation — not burnout, but a kind of buzzing. A fog that left them exhausted and struggling to concentrate.
“It was like I had a dozen browser tabs open in my head, all fighting for attention. I caught myself rereading the same stuff, second-guessing way more than usual, and getting weirdly impatient. My thinking wasn’t broken, just noisy — like mental static.”
— Senior engineering manager, BCG study participant
The study’s authors — including psychiatrist Gabriella Rosen Kellerman and BCG managing director Matthew Kropp — found that this mental strain carries real costs: increased errors, decision fatigue, and intention to quit.
The opposite of cognitive surrender
Kellerman, who also co-authored last year’s HBR report on “workslop” (AI-generated drivel that colleagues have to fix), draws a sharp distinction. Workslop reflects cognitive surrender — handing work to AI and not paying attention to the output. Brain fry is the opposite: trying to go “intelligence to intelligence” with the AI.
Sound familiar? If you’ve ever lost three hours trying to wrangle an agent that “almost” gets it right, you’ve been there.
Vibe coding paralysis
Francesco Bonacci, CEO of Cua AI, described the phenomenon as “vibe coding paralysis” in an essay on X:
“I end each day exhausted — not from the work itself, but from the managing of the work. Six worktrees open, four half-written features, two ‘quick fixes’ that spawned rabbit holes, and a growing sense that I’m losing the plot entirely.”
This maps almost exactly to what the OnTilt framework measures as Session Escalation (the “just one more fix” spiral) and Dark Flow (timeless absorption that feels productive but produces nothing).
Acute, not chronic
One critical finding: brain fry is acute, not chronic. Unlike burnout, which builds over time and degrades performance persistently, brain fry goes away when you take a break.
“When they take a break, it goes away,” Kellerman told CNN.
This distinction matters. It means the problem is solvable with behavioral design — structured breaks, session limits, cognitive boundaries. Exactly the kind of practices the OnTilt Checklist was built for.
What this means for the OnTilt framework
The BCG study validates what we’ve been mapping through a different lens. Their “AI brain fry” maps onto three of our six dimensions:
- Loss of Control — “Contrary to the promise of having more time to focus on meaningful work, juggling and multitasking can become the definitive features of working with AI” (Kellerman & Kropp)
- Session Escalation — The near-miss loop. “Almost works” becomes “just one more try” becomes three hours gone.
- Dark Flow — The buzzing. Absorption without agency. Hours vanish into managing, not creating.
Kropp compared the experience to “someone who just learned to drive being given a Ferrari. You can go really fast, but it’s easy to lose control.”
The six-worktree test
If you read Bonacci’s quote and thought “that’s me,” here’s a concrete check. At the end of your next AI-heavy day, count:
- How many contexts (worktrees, branches, chat threads) are open?
- How many of them produced a shipped result?
- How do you feel — energized or buzzing?
If the answer is “many, few, buzzing” — that’s brain fry. And you can do something about it.
Take the OnTilt Self-Check — 3 minutes, anonymous, designed to surface exactly these patterns.
Sources:
- Morrow, A. (2026, March 13). “AI is exhausting workers so much, researchers have dubbed the condition ‘AI brain fry.’” CNN Business Nightcap. Link
- Kellerman, G.R. & Kropp, M. (2026). “AI Brain Fry” study. Harvard Business Review / Boston Consulting Group.
- Bonacci, F. (2026). Essay on X regarding “vibe coding paralysis.”
OnTilt is a research project studying behavioral patterns in AI-assisted work. The quiz is a self-check tool, not a diagnostic instrument. Read more on our About page.