AI Brain Fry: When AI Doesn't Relieve, but Overwhelms
Why intensive AI use can cause mental exhaustion, based on March 2026 research by BCG and UC Riverside.
Lately, I have been seeing this more and more from developers on social media: frustration, exhaustion, and annoyance around AI.
It came up so often that I decided to look into it. Even though I am not a developer, I am convinced that knowledge workers will rely on AI even more heavily in the future. I wanted to understand what was behind this. Is there a real connection, or just coincidence? I found a clear signal.
AI Brain Fry - when AI doesn’t relieve, but overwhelms
A new study by BCG and UC Riverside (Harvard Business Review, March 2026) shows that intensive AI use can put people at risk of mental exhaustion - a phenomenon the researchers call AI Brain Fry.
One likely cause is multitasking while supervising AI activities.
Key findings from 1,488 surveyed US knowledge workers
- 14% report mental fog, difficulty concentrating, and headaches after intensive AI use.
- Most draining: overseeing multiple AI agents simultaneously (+12% more mental fatigue).
- Those affected make 39% more serious mistakes and show higher intention to quit.
- Most affected areas: Marketing, IT, Engineering, and Finance.
Interestingly, it is not AI skeptics who experience Brain Fry - it is early adopters, the people who use and monitor AI most intensively.
How Brain Fry shows up in everyday work
In most teams, this does not begin with a dramatic breakdown. It usually starts with smaller patterns that are easy to ignore:
- Decision quality drops even though more information is available.
- People spend more time supervising AI output than doing focused work.
- Context switching increases because multiple agents and tools run in parallel.
- A “productive” day still ends with mental fatigue and low clarity.
These are early warning signals. Ignoring them makes costly mistakes and team frustration more likely over time.
Why this happens
AI removes part of the work, often the part people find boring, but it also adds a layer of cognitive coordination: checking responses, comparing outputs, validating facts, and deciding what to trust. At the same time, this means more information to process.
In addition, tasks can often be completed faster at first. That tends to lead to more work packages, more task variety, and a broader scope of responsibility.
This points to a simple pattern: the more tools are used in parallel, the higher the risk of AI Brain Fry.
AI can help too
Burnout is often described as a more emotional strain, while AI Brain Fry is better described as cognitive overload.
AI can reduce emotional exhaustion by creating time for positive, restorative activities. But if that newly created time is used only for intensive supervision of multiple AI agents, it can still lead to mental exhaustion - and therefore to AI Brain Fry.
Final thought
If AI adoption is meant to be long-term, we need to manage cognitive load as seriously as productivity metrics. The goal should not be “more AI at any cost” - it should be better work systems where humans and AI each do what they do best.
One point is easy to forget: AI is simply another tool. The decisive difference does not come from the tool itself, but from our ability to use it in a meaningful way.
How does this show up in your own day-to-day work?