Three Hours with AI
A few months ago, I was writing a leadership report for a consulting client. Tight deadline, the kind that normally means a full day of research before you write a single sentence. I opened an AI assistant and started working.
Three hours later, the draft was done. Not a rough sketch. A complete draft with compelling structure, counterarguments I hadn’t thought to raise, insights I might have missed on my own. It was better than what I would have produced in a day, and it took a fraction of the time.
What AI Speed Feels Like
I kept scrolling through the report, half expecting to find the seams, the places where it would fall apart. It didn’t. The gap between idea and finished output had collapsed, and it felt like a superpower.
Then I closed the laptop. Something I didn’t expect: unease.
An awareness that I hadn’t struggled enough. The draft was good. I hadn’t wrestled with the material the way I normally do, hadn’t sat with confusion long enough for my own thinking to emerge.
I remembered my mother at the kitchen table. I was maybe ten, working through French homework. The conjugations wouldn’t stick. I wanted her to just tell me the answers. She wouldn’t. “Discover them yourself,” she said, “and the learning will stick.” I remember the frustration, the resentment, the stubborn belief that she was being unhelpful.
That memory surfaced uninvited. The parallel was uncomfortable. I had gotten the answers. I had skipped the discovery.
What AI Shortcuts Cost You
Nassim Nicholas Taleb’s concept of antifragility describes systems that grow stronger from stress. Muscles need resistance. Immune systems need exposure. Bones need impact. Remove the stress, and these systems don’t stay the same. They weaken.
The same principle applies to thinking. When I wrestle with a difficult brief, I’m not just producing a deliverable. I’m building the capacity to produce the next one. The confusion, the false starts, the moment where the structure finally clicks: these are not inefficiencies to be optimized away. They are the practice through which judgment develops.
This is what makes AI adoption that preserves capability so difficult. The thing you’re trying to protect is the thing the tool is designed to bypass.
A 2023 study by Fabrizio Dell’Acqua and colleagues at Wharton found that professionals using high-quality AI assistance showed reduced engagement with the work itself. They called it “falling asleep at the wheel.” The output looked fine. The thinking behind it had thinned. The professionals didn’t notice the shift.
The erosion is silent. You feel more productive, not less capable. The gap between those two experiences is where the real cost hides.
Organizations face this at scale. When entire teams default to AI for thinking work, the collective capacity for original thought doesn’t hold steady. It declines. Not dramatically. Gradually enough that nobody sounds the alarm.
The Kitchen Table, Again
I still use AI. I used it during the drafting of this piece. The question was never whether to use it. It was how to use it without losing what it so easily reveals about our own thinking.
My mother wasn’t against efficiency. She understood that the struggle wasn’t an obstacle to learning. It was the learning.
Her voice comes back to me when I work with AI now. Not as a memory I choose to revisit, but as a presence at the back of my mind, reminding me to preserve what was worth the effort. Choosing what not to automate turns out to be one of the most important decisions we face.
The answer isn’t always obvious. Sometimes the shortcut is the right call. But knowing when it isn’t requires the kind of judgment that only develops through effort.