What happens when the tool that's supposed to make you faster actually burns out your brain?
Boston Consulting Group surveyed 1,488 full-time U.S. workers and found that 14% of those using AI tools regularly experience what the researchers call "brain fry" - a specific type of mental exhaustion distinct from regular burnout. It hits attention and working memory hardest, and the symptoms are concrete: mental fog, difficulty concentrating, slower decisions, and a buzzing sensation that doesn't go away after a coffee break.
The numbers behind it are worse than the label suggests.
The Error Rate Problem
Workers experiencing brain fry reported 11% more minor errors and 39% more major errors compared to their peers. They also showed 33% greater decision fatigue and 12% higher mental fatigue scores. Most concerning for employers: heavy AI users with brain fry were 39% more likely to say they intended to quit.
Marketing departments got hit hardest at 26%. Engineering, IT, finance, and operations all showed elevated rates. Legal came in lowest at 6%, which makes sense given that legal work tends to involve careful, sequential analysis rather than the rapid context-switching that triggers cognitive overload.
The Three-Tool Cliff
One of the study's most practical findings: productivity peaks when workers use three AI tools simultaneously and drops off sharply after that. This matches what most people who juggle ChatGPT, a coding assistant, and an AI writing tool already feel intuitively. Adding a fourth or fifth tool doesn't make you more productive. It makes you a project manager for a team of AI assistants, each requiring oversight, correction, and context.
The study specifically calls out AI oversight requirements as the most taxing activity. It's not using AI that fries your brain. It's monitoring AI output, catching its mistakes, and deciding when to intervene. Software developers are particularly affected because AI coding agents can generate enormous volumes of code that still needs human review.
What Actually Reduces the Load
The BCG data points to some straightforward interventions. Manager support (just answering questions about AI tools) cut mental fatigue by 15%. Clear messaging about work-life boundaries reduced it by 28%. Using AI specifically to eliminate repetitive, routine tasks lowered burnout by 15%.
The flip side: workers left to figure out AI tools on their own without organizational support saw fatigue increase by 5%. Teams with inconsistent AI adoption, where each person uses different tools differently, experienced more strain than teams with coordinated approaches.
None of this means you should stop using AI tools. But if you've noticed that your afternoon focus has gotten worse since you started running three AI assistants at once, the BCG data suggests you're not imagining it. The cognitive cost of supervising AI output is real, measurable, and - for about one in seven workers - significant enough to affect both error rates and job satisfaction.