Everyone says AI makes you “10x more productive.” I’m not sure about that. What it actually made me is… more deliberate.
When execution is basically free, the bottleneck shifts. It’s no longer can I build this? It’s should I build this?
That sounds obvious, but most of us (me included) are terrible at it. We confuse motion for progress. And AI just cranks the treadmill speed up to 11.
Here’s what I mean.
The other weekend, I let myself play around with the idea of “AI-generated developer dashboards.” Normally, something like that would eat a week of evenings. This time, I had three versions running before breakfast: a React prototype, a Python backend spitting out metrics, and a half-decent mock landing page.
Impressive? Maybe. Useful? Not really. By Sunday night I realized I’d basically built three beautifully useless toys. Execution had been trivial. The problem was never execution—it was me chasing shiny objects.
That’s the AI paradox. It lowers the cost of building so much that the real scarcity becomes taste. Judgment. The ability to say no.
Because here’s the dark side: the opportunity cost of distraction just went up. Before, if I burned a week tinkering on something dumb, at least I learned a few low-level tricks. Now I can burn a week and end up with a full microservice, a CI/CD pipeline, and a Terraform config… for an idea that didn’t deserve any of it. Congratulations, I’ve industrialized my dead ends.
I’ve caught myself doing this with infrastructure experiments, too. AI will happily generate Kubernetes manifests, Helm charts, and CI workflows for whatever hair-brained service I throw at it. The code even looks plausible at first glance. Then I deploy it, watch it explode, and realize the whole thing never needed to exist in the first place. It’s the most polished waste of time imaginable.
And this is why restraint has suddenly become a superpower. The real work isn’t generating more; it’s filtering harder. AI will give you 50 rabbit holes before lunch. If you’re not ruthless about which one you go down, you’re just automating your own distraction.
The old mantra was “ship fast and break things.” AI makes that easier than ever. But there’s a hidden multiplier effect: fast execution with bad strategy doesn’t just fail—it fails louder. You don’t just waste time, you waste time at scale. Meanwhile, the teams with clear strategy and discipline can use the exact same tools to compound wins. Same technology, wildly different outcomes.
This is why I think “thinking” has quietly become underrated. Tinkering used to be the path to learning. Now tinkering is dangerous. You can dig a perfect hole in the wrong place faster than ever. Spending more time deciding where to dig—that’s the skill worth leveling up.
Developers don’t usually like to hear that. We want to build. But in an AI-first world, the rarest and most valuable act might be… not building. Closing the tab. Saying no to the prototype. Choosing boredom over the dopamine hit of “look what I got running.”
So no, AI didn’t make me more productive. It made me picky. It forced me to care about what I was building in the first place.
And that’s the paradox: AI made execution trivial, so the premium is now on taste, judgment, and focus.
If you can’t decide what matters, AI will happily help you drown in what doesn’t.