Yes, I Use AI. I Also Use My Brain.

People sometimes ask me how I use AI. The shortest answer is:

I use it to think.

I also use it to write, edit, argue, brainstorm, research, reorganize, test language, notice contradictions, and occasionally generate something so spectacularly overconfident that I have to laugh.

But mostly, I use it to think.

Generative AI is extraordinarily good at pattern recognition. Human brains recognize patterns too, although obviously not in the same way. We compare. Associate. Categorize. Remember. Notice echoes. Build stories from fragments. Ask: What does this remind me of?

So sometimes I give a machine something I have made and ask:

What do you see?

Not: Tell me this is brilliant. Not: Tell me whether I am brilliant.

Those are terrible questions for a machine that is remarkably capable of telling a human being what she appears to want to hear.

I mean something more interesting.

What patterns do you detect here that I might not see from inside the work?

That was the question behind an experiment I recently ran with my book, A Trial of Color. I asked AI what books it would put mine beside. Its answer wandered through Joan Didion, Esmé Weijun Wang, Susannah Cahalan, Oliver Sacks, Maggie Nelson, Robert Pirsig, and even the literature of judgment and cognitive error.

That was flattering. But the flattery wasn’t the interesting part. The pattern was.

It saw grief. It saw psychiatric self-investigation. It saw disability and perception. It saw epistemology. It saw a courtroom instinct: evidence, testimony, reconstruction, competing explanations. It saw a person trying to understand what happens when the mind encounters a system capable of generating language, meaning, connection, and apparent recognition at extraordinary speed.

And then it produced a phrase I hadn’t given it: AI memoir as primary-source evidence.

Huh. That interested me. From inside the manuscript, all of those subjects feel connected because I lived the connections. But writers have a peculiar problem. We know our own work too well. Eventually the structure becomes almost invisible to us because we are standing inside it.

So I sometimes use AI the way I might hold a piece of writing up to a very strange mirror: tell me what shape you see. That doesn’t mean the mirror gets to decide what the object is.