Marketing is leading the pack — the wrong way.
New HBR research on 1,488 full-time workers found that marketers report the highest rate of "AI brain fry" of any function. 26% say they're cognitively exhausted from supervising AI output — more than engineering, finance, legal, or IT.
That's the bad news.
The good news: the same study drew a clean line between AI that makes work worse and AI that makes work better. And it has nothing to do with which model you're using.
Oversight vs. toil — the distinction that matters
The authors separate AI deployments into two categories.
AI that demands constant oversight. Drafts copy you have to edit. Proposes decisions you have to review. Runs workflows that need a human rubber stamp at every turn. Top users of this kind of AI report 12% more mental fatigue, 19% more information overload, and 39% higher intent to quit.
AI that absorbs toil. Handles list uploads. Routes leads. Reconciles duplicates. Formats reports. The grinding, repetitive work nobody actually wanted to do. Among workers whose AI replaces this kind of work, burnout drops 15%.
Same technology. Wildly different outcomes. The difference is what you ask it to do.
This matches what we hear from MOps teams every day
When Mary — our AI MOps teammate — takes a list upload off someone's plate, the feedback isn't "I'm more productive." It's "I actually like my job again."
Their time goes to campaign strategy, creative, and the kind of analysis that moves the business. Not to cleaning columns in a spreadsheet at 9pm on a Tuesday.
The HBR authors put it well: AI helps most when it replaces tasks to give people time for "restorative, positive activities."
In MOps, that means automating the toil — not layering more agents on top of already overloaded marketers and calling it transformation.
How to tell which kind of AI you're building
The measurement you choose reveals the strategy.
If your AI strategy is measured in tokens consumed, agents orchestrated, or prompts shipped, you're probably building brain fry. You're adding supervision surface area, not removing it.
If it's measured in hours given back and toil eliminated, you're building the opposite. The work gets quieter. The team gets their afternoons back. Intent-to-quit goes down, not up.
The question isn't "how much AI are we using?" The question is "what did our marketers do with the hours AI gave them?"
If the answer is "reviewed more AI output," try again.
If it's "ran a better campaign, talked to a customer, made a creative decision" — you're on the right side of the line.
Source: When Using AI Leads to Brain Fry, Harvard Business Review, March 2026.
