Jackson Bates

Writing Is Thinking

The sculpture is already in the marble. That's Michelangelo's idea, anyway. You don't create it - you remove everything that isn't it. Chip away the unnecessary, the untrue, the almost-right. What remains was always there. You just had to find it.

Writing works the same way. Or it's supposed to, anyway.

The word "essay" comes from the French essai - Montaigne's term for his own writing. It means attempt. Trial. We use the word assay now in English, too - to test the integrity of something. An essay represents a thinking-out-loud rather than a conclusion delivered. The essay was never meant to be a finished argument. It was the record of someone working one out.

I spent years watching students discover this under duress. Timed conditions, handwritten, no way to copy-paste chunks or move ideas around once they were on the page. When a better thought arrived late - the one that actually complicated something, that pushed back against what they'd already written - they had nowhere obvious to put it. So they squeezed it wherever it would fit. Sentences jammed between the lines already written. Ideas snaking up margins looking for breathing room. Arrows connecting afterthoughts to paragraphs written twenty minutes earlier.

Messy. Hard to read. Almost always the parts worth decoding.

The early paragraphs were the easy thoughts - the ones already formed before the pen touched the paper. The margin notes were where the real thinking broke through. The marble giving way.

I also watched what happened when students made mistakes. The correct response is a single clean line through the unforgivable sin error. Move on. But some students would obliterate theirs - aggressive scribbles, heavy scratching, paper sometimes tearing. Nothing legible left.

My hunch was always that it wasn't about hiding the mistake from me. It was about hiding it from themselves. What you write carries something of you in it - it came from your thinking, which means it came from you. Seeing something you produced that is plainly wrong can feel like an affront. Not just to the work. To you.

Writing externalises your thinking in a way that makes it visible. And visible means vulnerable.

A colleague mentioned recently, pretty casually, that they use AI for all of their writing. I didn't say anything at the time. But it's been sitting with me since.

Because here's the thing. Writing has two purposes. There's the outward one - communication, getting an idea across. That's the part most people think about. But there's an inward purpose too, and I think it's the one that gets skipped without anyone noticing.

An LLM can play almost any role in the writing process. Planning partner, outline generator, first draft, editor, sin eater, proofreader. None of those handoffs is inherently wrong. But each one is an opportunity to think - or an opportunity to have the thinking done for you. The problem isn't the handoff. It's not noticing you've made it.

When you're looking at output on a screen that reflects your prompt, is roughly in your voice, feels familiar - it's easy to experience that as your thinking. The test worth applying is: how often are you pushing back? How often are you telling it that something's missing, that the nuance is wrong, that it hasn't quite captured the spirit of what you meant? If the answer is rarely - if you're mostly reading and nodding - you're probably rubber-stamping something plausible rather than arriving at something true.

Maybe the real danger of generative AI isn't that it chips away at the marble for you until David emerges.

It's that it gives you more marble.