Speed of AI adoption is not quality of adaptation

Once again about AI adoption — what people are actually adapting into.

I came to LinkedIn to find people who think about AI's future the way I do — not through hype cycles, but as a long-term global shift that will reshape our day-to-day life: how we study, how we work, how we process information.

Some time ago, I crossed paths with Adi Stan in a comment thread, and he suggested I check out his research paper "The Age of Cognitive Divergence". I did. And the way he frames AI adoption genuinely extended how I think about this whole situation.

The core thesis: AI adoption doesn't democratize intelligence. It stratifies it.

L1 (Passengers)

dependent users stuck in what Stan calls the "uncertainty-reassurance cycle". They ask AI, accept the first answer, and move on—cognitive atrophy in slow motion.

L2 (Operators)

the ones who use AI to do the same things faster. Linear gains. But here's the catch: they're sitting in the maximum automation risk zone. The "safe middle" isn't so safe.

L3 (Architects)

a small group entering actual symbiosis with AI. Exponential productivity. Not because they prompt better, but because they think through the tool.

The problem: L1 and L2 feel productive. But one is building dependency, the other is sitting in the automation kill zone. Only L3 is actually compounding.

This sharpens something I haven’t thought about yet. 🎯

"Fast adaptation" means nothing if it's adaptation into dependency, not into thinking.

An engineer who learns to prompt AI in a week but never questions the output has adapted fast — straight into "Passengers". An engineer who takes longer but learns to iterate, verify, and think through the tool — that's the one who'll turn AI into real career leverage.

Speed of adoption ≠ quality of adaptation.

So, fellow CTOs and CEOs — are we even measuring the right thing?

How fast do people use AI, or how well do they think with it? 🧠

Curious what you're seeing in your teams.

👇 Link to Adi's paper in the comments.