AI brain fry: what AI fatigue is, and what helps

In Dr Prompt’s waiting room, the complaint is always the same: a head that buzzes after a day spent with AI agents. Researchers now have a name for it.

Tokens Anonymous is a parody. This page isn’t: every figure below comes from a cited study. It is not medical advice.

What is AI brain fry?

“AI brain fry” is the name researchers at BCG gave to “mental fatigue from excessive use or oversight of AI tools beyond one’s cognitive capacity”⁷. It comes from a workplace study: it is not a medical diagnosis.

They set it apart from burnout. Burnout is chronic workplace stress, more emotional; brain fry is acute, and cognitive⁷. So if you searched for “AI fatigue”, this is the fatigue of working with AI, not the fatigue of hearing about it.

The signs

Participants in the study described⁷:

• a “buzzing” feeling, or a mental fog;
• difficulty focusing, and slower decision-making;
• headaches;
• after intensive back-and-forth with the tools, an inability to think clearly, “like a mental hangover”;
• for several of them, the need to physically step away from the computer to “reset”.

One senior engineering manager summed it up:

“I was working harder to manage the tools than to actually solve the problem.”⁷

What the study found

BCG surveyed 1,488 full-time US workers at large companies, across industries, roles and levels. The results were published in Harvard Business Review in March 2026, and every figure in this section comes from there⁷.

• How common it is. 14% of participants who use AI for work said they had experienced it: from 6% in legal roles up to 26% in marketing.
• Oversight is the main driver. When their AI work required a high degree of oversight, workers spent 14% more mental effort, and reported 12% more mental fatigue and 19% more information overload. A workload that AI tools had made heavier also predicted mental fatigue.
• Three tools, then it dips. Perceived productivity rose from one AI tool to two, rose less with a third, then dipped beyond three tools used at the same time.
• What it costs. People with brain fry reported 33% more decision fatigue, and scored 11% and 39% higher on minor and major error frequency. 34% of them actively intended to quit, against 25% of the others.
• It isn’t burnout. Using AI to cut routine, repetitive tasks went with 15% lower burnout scores, but not with lower mental fatigue scores. AI can ease burnout when it removes toil, and wear you out when it demands constant oversight.

The limits: everything comes from what participants reported in a survey, and the authors work at BCG, a consulting firm that advises companies on AI⁷.

What helps

Dr Prompt usually prescribes another prompt. The study points elsewhere⁷:

• Fewer agents at once. Productivity dipped beyond three tools used simultaneously. The authors call for limits on how many agents one person oversees, the way there are spans of control for managers.
• Give AI the toil, not your attention. Offloading repetitive tasks went with less burnout; constant oversight went with more fatigue.
• Stop iterating sooner. In the authors’ words: “Just because a worker can keep iterating with AI at a low marginal cost does not mean they should.” They recommend building skills such as problem framing, planning and prioritization.
• Step away. Several participants had to leave their computer to “reset”. A finance director put the work off until the next day, “when I could think”.
• If you manage people. Answer their questions about AI: their mental fatigue scores were 15% lower, and 5% higher when managers expected them to figure it out alone. Say what AI changes about workload: expecting more work because of AI went with 12% higher fatigue. Reward impact, not usage or token consumption. Employees who felt their organization values work-life balance had 28% lower mental fatigue scores.

Brain fry, AI addiction and ADHD

Brain fry is about fatigue. The rest of this site is about the other half of the problem: why it’s so hard to stop. An agent that sometimes nails it and sometimes fails, with no warning, is variable-ratio reinforcement, the mechanism that makes slot machines so hard to walk away from; well-known developers describe exactly this¹. In young adults with ADHD, it’s reward processing, not attention deficits, that predicts internet addiction². And “AI addiction” is not a recognized diagnosis: some researchers question the concept entirely⁵.

Take Dr Prompt’s test (for laughs), or answer thereal anonymous survey: if you use AI for work, it asks how many agents you ran in parallel, how tired you’ve been at the end of the workday and how hard it was to concentrate.

Sources

1. One More Prompt: The Dopamine Trap of Agentic Coding, Quentin Rousseau, March 2026 (quotes Steve Yegge and Garry Tan)
2. Motivational but not executive dysfunction in ADHD predicts internet addiction, Psychiatry Research, 2020
5. People are not becoming “AIholic”: Questioning the “ChatGPT addiction” construct, Addictive Behaviors, 2025
7. When Using AI Leads to “Brain Fry”, BCG, Harvard Business Review, March 2026 (1,488 US workers)

Back to Dr Prompt