- Apr 28
- 2 min read
Congratulations. You’ve been promoted.
You’re now middle management.
The team you’re managing? Claude, ChatGPT, Copilot, whatever AI tool your company has rolled out, and possibly two more depending on the day. They’re brilliant. They’re tireless. They will not stop pinging you for feedback, review, correction, and the occasional reprompt.
Welcome to your unpaid promotion.
Last week on How I Work I sat down with Gabriella Rosen Kellerman, a behavioural scientist at BCG whose recent HBR study has been doing the rounds. She’s put a name to something heaps of us have been feeling but couldn’t quite articulate. It’s called AI Brain Fry.
Gabriella’s definition: mental fatigue from excessive use or oversight of AI tools beyond your cognitive capacity.
Brain fry isn’t burnout. It sits in its own category. Doesn’t increase burnout. Doesn’t decrease it. Which means a lot of us are walking around fried, and the standard wellbeing surveys are not catching it.
The cause is oversight. The cognitive load of managing our AI chat bots.
When AI handles a repetitive task you didn’t want to do anyway, burnout actually goes down. Lovely. But when you’re orchestrating tools, prompting, reviewing, correcting, deciding what’s good enough, fixing what isn’t, you’ve taken on a whole new job description.
Time back, in practice, looks suspiciously like direct reports.
A few signals you’re in the Brain Fry zone:
You’ve got more than three AI tabs open right now. (Gabriella’s research found productivity peaks at 2 to 3 tools and tanks after that.)
You’re switching between agents and chats/threads constantly while one of them finishes processing.
Your role is operationally or technically heavy. (Marketing topped the brain fry charts at 26%.)
You finished work yesterday with a kind of cognitive static you couldn’t name.
Sound familiar? Probably.
So what can we do to stop our brains from frying?
1. Notice the fry as it lands. Gabriella shared this analogy: When you go for a run, you learn your breath. You feel when to slow down. We need the same awareness with AI. Build a “how’s my brain?” check into your day. Mid-afternoon is usually when the static kicks in (Gabriella’s non-negotiable: 24 hours digital-free, once a week, helps too).
2. Cap your tools. Two or three is the sweet spot. Anyone running 25 tabs using 8 different tools at once is doing performance art.
3. Aim AI at the work you don’t want to do. This was the one pattern in the study that actually reduced burnout. Workers who used AI to remove repetitive tasks they didn’t enjoy reported 15% lower burnout. The win comes from getting time back. The fry comes from filling that time with three more agents to oversee.
4. Stop measuring volume. Start measuring value. Meta currently counts AI-generated lines of code as a performance metric. Some companies reward token consumption. Predictable result: people use AI more, produce more slop, feel more fried. If you lead a team, pick the outcome that actually matters to the business and measure that. Activity will sort itself out underneath.
5. Manage your new team properly. Because that’s the actual job now.
Until next week,
Amantha