AI Brain Fry Is Real. Here’s What the Research Says About Protecting Your Team From It.
New research from MIT, UC Berkeley, and a dozen other institutions is telling us something most leaders don't want to hear: the people using AI the most may be the ones losing the most. Here's what's actually happening — and what you can do about it.
Field Notes from ArtWorks — July 2026
There is a term making the rounds in cognitive science right now: AI brain fry.
It is not a metaphor. Researchers from ActivTrak analyzed the digital activity of more than 10,000 workers and found that when people adopted AI tools, their work life became more intense, not less. The time early adopters spent on email, messaging, and chat more than doubled. Use of business software rose by 94 percent. Meanwhile, the time people spent on focused, uninterrupted work fell by 9 percent.
They got busier. They got less deep.
This is not what anyone was promised.
The three things happening inside your organization right now
A team of researchers at MIT's Media Lab found that people's brain connectivity declines by as much as 55 percent when using AI compared to performing similar tasks without it. Gamma-wave activity — a measurable sign of cognitive effort — dropped by roughly 40 percent during AI-assisted work. A study published by SBS Swiss Business School found a significant negative correlation between frequent AI tool usage and critical thinking abilities.
In a widely cited study, researchers put subjects through 10 minutes of AI-assisted problem solving, then took the AI away. Those who had used AI performed worse and gave up more frequently than people who had never used it at all.
Ten minutes. That's all it took.
UC Berkeley's Haas School of Business found something equally interesting from the other direction: when using AI, workers started taking on tasks they had previously outsourced, because activities like coding and research became easier to attempt. They worked in the evenings, on weekends, in waiting rooms. They multitasked across multiple AI tools simultaneously. They produced more — and retained less.
Writer David Brooks, synthesizing this body of research in The Atlantic, identifies three emerging categories of knowledge worker in the age of AI. The first group uses AI to think less — more productive in the short term, measurably hollower over time. The second group intends to use AI thoughtfully but gets seduced by its efficiency, gradually losing the cognitive habits they meant to protect. The third group — the smallest — uses AI as a scaffold to think harder and go further.
The unsettling part of Brooks' analysis is his prediction: the gap between these three groups will widen. Fast. He calls it cognitive polarization — a divergence that may ultimately matter more than economic inequality or political polarization in shaping who thrives and who doesn't in the decades ahead.
Most organizations have no idea which group their people fall into.
The volition problem
Brooks' central argument is worth sitting with: when intelligence is plentiful, volition is valuable.
AI does not hunger. It does not have ambitions or wounds or a self that is trying to become something. It predicts. It synthesizes. It optimizes. What it cannot do is want things — and wanting things, it turns out, is most of what drives human performance at the highest level.
The people who will thrive in an AI-saturated world are not necessarily the smartest. They are the ones with the strongest relationship to effort — the ones who seek cognitive complexity rather than avoid it, who treat difficulty as the point rather than the obstacle.
The question this raises for every leader is not: Are we using AI effectively?
It is: Are we building the conditions that cultivate volition — or quietly depleting them?
That distinction matters because volition is not fixed. It is extraordinarily sensitive to context. Self-determination theory — the most robust framework we have for understanding human motivation — identifies three conditions that sustain it: autonomy (I am in control of my choices), competence (I am developing my skills), and relatedness (people here care about me).
When those conditions are present, people develop. They take on difficult problems. They build the judgment required to supervise AI rather than be replaced by it. They become the third category in Brooks' taxonomy — the ones who use intelligence tools to go further, not the ones who use them to coast.
When those conditions are absent — when people feel surveilled rather than trusted, stalled rather than developing, invisible rather than seen — something else happens. The optimization mindset takes over. Output becomes the measure. Quality becomes someone else's problem. The AI fills the gap where engagement used to be.
And the organization doesn't see it until people start leaving — or until the work starts failing in ways that are hard to trace back to their source.
What you can actually do about it
The research is not without practical direction. Brooks identifies several techniques that the high-performing group uses to maintain cognitive depth while benefiting from AI:
Start with a blank page before engaging the tool. Write your own analysis first. Then ask AI to challenge your thinking — not produce it.
Ask for hints, not answers. People who ask AI for background context or clarifying frameworks maintain motivation and ability. People who ask it for conclusions lose both.
Rotate tasks. Every AI-assisted task should be followed by one that requires unassisted thinking. The cognitive muscle atrophies quickly and comes back slowly.
Make a sharp distinction between rote work and creative work. AI writes the functional email. Humans write the memo that matters.
These are individual practices. But the more important lever is organizational.
The research is consistent on this point: motivation, curiosity, and cognitive depth are not intrinsic properties that people either have or don't. They are responses to environment. A person who feels genuinely seen, whose development is treated as a priority, whose work connects to something they actually care about — that person is not going to surrender their judgment to a chatbot. They have too much invested in using it well.
The person who feels depleted, uncertain about their future, disconnected from the people around them — that person is exactly who the optimization mindset is waiting for.
The measurement gap. Again.
This is the part that keeps coming up in the research.
Organizations are measuring AI adoption. They are measuring productivity. They are measuring outputs. What they are not measuring — what almost no one is measuring — is the human conditions that determine whether any of those outputs are sustainable.
Are your people developing or just producing? Are they building judgment or offloading it? Are they in the conditions that cultivate volition — autonomy, competence, relatedness — or quietly losing the capacity to do hard things without a prompt?
The ActivTrak data tells you how many hours people spent in business software. The MIT data tells you what happened to their brain while they were doing it. Neither tells you what's actually happening for the people inside your organization right now — what's driving them, what's blocking them, what they need that no one has thought to ask about.
That's the gap. And in an era when the difference between the people who use AI to go further and the people who use it to coast is going to compound dramatically, that gap is no longer a soft HR concern.
It is a business risk.
The question worth asking your team this week
Not: Are you using AI?
Not: How much time are you saving?
But: Are you still doing the hard thing? Are you still in the work — or are you supervising it from a distance?
The organizations that will navigate the next decade well are not the ones that deployed AI first. They are the ones that kept asking what was happening to the people inside them — and had the tools to hear an honest answer.
That's the work. The rest is software.
ArtWorks is a leadership coaching and organizational development practice based in Dallas, TX. Interplay™ is our human flourishing diagnostic — built on Harvard's Human Flourishing Framework, deployed across eight institutions, and available for workplace, higher education, and career applications. If something in this piece landed, let's talk.
Field Notes from ArtWorks is our institutional research and analysis series — grounded in the latest thinking on leadership, human flourishing, and organizational change.
