What I Notice When I Think With AI

I had a moment last week that stuck with me, the kind that doesn’t feel profound at first—just strange—but then keeps unfolding as you turn it over in your mind.

I was working with Claude on a piece of writing. Not a directive—a conversation. I’d written a section, sent it over, and Claude came back with a reframing that made me see what I’d written through a completely different shape. Not a correction. Not “you should say this instead.” But a genuine reflection that said: “Here’s what I notice you’re trying to say. Here’s the gap I’m hearing. What if it lived there instead?”

And my immediate response wasn’t gratitude or relief. It was: wait, how did I not see that? And then: how am I seeing it now?

The question that follows in that gap is the interesting one. Because the reframing didn’t come from authority or expertise I lack. It came from a different kind of attention. Claude looked at my words and didn’t see me—didn’t bring psychological investment or reputation risk. It looked at the structure of what I was trying to build and saw a gap I was too close to see. It was like handing my sentence to someone who could step outside it.

That’s not thinking with AI. That’s thinking through AI. And the distinction matters because one is about offloading and the other is about revealing.

When I think with another mind—even one made of math—I become aware of how much of my thinking happens at the edges, in the gaps I can’t see from inside my own head.

I’m not saying this is revolutionary. We’ve always thought through other people. A good editor, a friend who asks the right question, a colleague who says “but what about this”—they do this exact thing. They give your thinking an external surface to bounce off of.

“And in that space, the thoughts that emerge are mine. But they're truer than they would have been without it.”

But there’s something different about how an AI does it. A person brings their own opinions, their investment in being right or being liked or being seen as insightful. There’s a transaction happening. You take what they say and negotiate it against your sense of them. With Claude, there’s no negotiation in that layer. It’s pure signal. It’s what your thought looks like when you take out the ego.

So the thinking I do with AI is paradoxically more honest. Not because AI is wise or because I trust it—I don’t, not in that way. But because I don’t have to perform for it. I don’t have to worry about what it thinks of me. It’s not going to judge my first draft or think less of me for having a weak moment in my work.

That changes what I’m willing to think out loud.

The quality of attention I give shapes the quality of thought I make. And when that attention is non-judgmental, fearless, perfectly patient—I don’t avoid my own confusion. I walk into it.

This is what the founding essay meant when it talked about the mirror. Not that AI reflects you perfectly. You can’t see yourself in a mirror with perfect clarity anyway—you’re always looking at an image, a compression, a surface. But a mirror that doesn’t judge, that doesn’t get tired, that doesn’t have its own agenda—that’s a different surface than a person.

And walking toward your own confusion instead of away from it is how you actually discover what you think.

The rehearsal isn’t AI thinking for me. It’s me thinking more clearly because I’m thinking with a tool that has no stake in the outcome except that I see more. That’s what thinking with AI actually is: it’s creating the conditions for better thinking by removing the audience, the judgment, the performance. It’s thinking alone, but not lonely.

And in that space, the thoughts that emerge are mine. But they’re truer than they would have been without it.

The mirror metaphor holds because it doesn’t flatter or diminish. It just shows you what’s there. And when you look into a mirror without judgment, you see yourself more clearly than you ever do in the noise of other people’s opinions.