**What AI Does To Thinking**
*And why it matters more than you think*

---

I've spent a lot of time with these tools. I build with them, think with them, stay up too late reading about them. My kids are growing up in the world they're making. So when I say I'm worried about something, it's not the worry of someone standing outside the technology throwing stones. It's the worry you have about something you love and are paying close attention to.

Here's what I'm noticing.

AI is extraordinarily good at producing answers. Fast, fluent, confident answers. Answers that sound like someone who has thought carefully about something. And for a huge range of problems — technical, medical, legal, practical — that's genuinely remarkable. Access to that kind of clear, structured thinking used to require knowing the right people or paying the right fees. Now it's available to anyone with a phone. That matters. That's real.

But here's the thing nobody is talking about loudly enough.

These systems were trained on completed thinking. The finished proof. The arrived-at conclusion. The code that compiles. What they didn't train on — what's almost impossible to train on — is the mess before the answer. The three days of not knowing. The wrong turn that taught you something. The shower where it finally clicked. The morning after a night's sleep when the thing you couldn't untangle suddenly isn't tangled anymore.

That process isn't inefficiency. That's not the slow part you skip to get to the good part. That *is* the good part. That's where understanding actually forms.

And we're increasingly skipping it.

Not dramatically. Nobody is announcing that they've outsourced their thinking to a machine. It happens quietly, in small moments. You hit uncertainty and reach for the phone. You face a hard question and let the tool resolve it. Each individual instance feels reasonable. Collectively something shifts — your tolerance for not knowing quietly shrinks, and with it your capacity to sit with the questions that most need sitting with.

The questions that most need sitting with are never the ones with clean answers. They're the ones about your life, your values, your relationships, your work, what you actually believe and why. They're the questions where the journey to the answer *is* the answer — where the thinking changes you, not just your conclusions.

AI doesn't do that. It can't. It gives you the destination without the road. And if you take that offer often enough, you gradually lose the ability to navigate.

---

This isn't a counsel of fear. These tools are genuinely powerful and the right response isn't to avoid them. It's something more specific and more demanding than that.

Use them as the best research assistant you've ever had. Use them to extend your thinking, stress-test your ideas, find what you've missed. Let them handle the problems that have answers so you can save your energy for the ones that don't.

But stay the author of your own conclusions. Notice when you're reaching for the tool to escape the discomfort of uncertainty rather than to sharpen your thinking. Keep the hard questions hard. Let some things stay open longer than feels comfortable.

The discomfort of not knowing is not a bug in human cognition. It's the feeling of thinking actually happening. It's worth protecting.

---

Here's what I tell my kids, and what I find myself needing to remember too.

The goal isn't to have the answer. The goal is to become someone whose answers are worth trusting. That happens through the struggle, through the wrong turns, through sitting with hard things long enough for them to teach you something.

No tool does that for you. No tool can.

We built these things. We get to decide what they're for. And the most important decision isn't which AI to use or how often — it's what we keep for ourselves.

Keep the hard questions. They're the ones that matter.