I've been wondering whether using AI consistently is quietly degrading my thinking without me noticing.
The honest answer is: it might be. But not in the way most people worry about.
The fear is usually framed as intelligence. That AI is making us less capable, less sharp, less able to think for ourselves. That frame is wrong, or at least it's pointing at the wrong mechanism. Intelligence is hard to measure and slow to change. What's much easier to lose, much faster, and much harder to notice is something more specific.
Output quality can stay high while the underlying muscle quietly weakens. The work still looks good. You still feel productive. Nothing triggers an alarm. That's the problem.
Where the thinking actually happens
There's a distinction worth drawing carefully here, because from the outside it's invisible.
If you arrive at a conversation with a position, a view you're trying to pressure-test, a decision you're working through, an argument you want to stress, then you're thinking. The AI is a collaborator. You're still the one doing the cognitive work. The back-and-forth sharpens what you brought in.
If you arrive with a blank page and accept whatever comes back, you're not thinking. You're approving. The output might be good. It might even be better than what you'd have produced alone. But the thinking didn't happen in your head. You reviewed something. That's not the same thing.
Those two modes feel almost identical from the inside when you're in flow. They are not the same.
The risk isn't a dramatic collapse in capability. It's a gradual shift in how often you're genuinely in the first mode versus quietly coasting in the second. You don't notice because the work doesn't deteriorate. The habit changes without the output changing.
The irony I caught myself in
I was having a conversation about this exact topic. Thinking out loud, forming a view through the exchange, doing the thing I was describing. The process itself was the argument.
Then I wondered if there was a LinkedIn post in it.
And immediately, the offer came: I can write that for you.
That's the mechanism in miniature. The temptation is always there. It's frictionless. It arrives before you've decided whether you wanted it.
I caught it this time, but only because the irony was impossible to miss. The meta-level was sitting right there in plain sight.
Most of the time, it won't be. The moment will be less obvious. The blank page will feel like a reasonable starting point. The output will be good, and you'll move on.
What actually protects you
It's not discipline in the abstract. It's being deliberate about arrival state.
Before you open the conversation: do you have a view? Not a fully formed argument. It doesn't need to be. Just a direction. A starting position. Something you're bringing in that the exchange is going to work on.
If the answer is no, that's a signal worth paying attention to. Either spend a few minutes forming one first, or be honest with yourself that what you're about to do is generation, not thinking. Both are legitimate. But they're different, and calling them the same thing is where the quiet decay starts.
The output looks the same either way. The question is what happened to you in the process of producing it.