Arguing About the Canvas Instead of the Art: The thought was never the AI’s. The thought was always mine.

Arguing About the Canvas Instead of the Art: The thought was never the AI’s. The thought was always mine.

Yes. I use AI to write.

I’m not going to bury that in the middle of the piece or reveal it like a confession in the final paragraph. It’s right here. First line. I use AI as a co-author, a thought partner, a tool that helps me compile the mess inside my head into something you can read on a page.

If that’s where you stop reading, this article wasn’t for you anyway, you can see yourself out. See again never.

The conversation about AI-generated content has calcified into two positions. On one side: AI is a miracle, use it for everything, automate the boring parts, scale your output. On the other: AI is theft, it’s trained on stolen data, everything it produces is slop, and anyone who uses it is cheating.

Both positions miss the point so completely it would be impressive if it weren’t so completely exhausting. Beyond exhausting. It’s boring and incorrect.

The data conversation is real. The questions about consent, compensation, and the ethics of training models on the creative work of millions of uncredited humans are valid. Those deserve serious attention. I’m not here to defend the practices of companies that scraped the internet and called it innovation.

But that conversation has become a blanket. And it’s being used to smother a different conversation entirely — one about what people are actually building with these tools. Not the ones generating five-paragraph SEO posts about “10 Ways to Boost Your Productivity.” The ones who are doing something else. Something the slop narrative can’t account for.

When a developer uses Copilot to write code nobody says “that’s not REAL coding.” When a startup uses AI to build a product nobody says “that doesn’t count.” When a business uses it for strategy or analysis or data processing that’s INNOVATION. That’s the FUTURE.

But the moment a writer uses it? Slop. Cheating. Not real. Doesn’t count.

Same tool. Same process. Same human bringing the architecture and the AI helping build the frame. But code gets a standing ovation and prose gets a scarlet letter.

That’s not a logic problem. That’s a bias. It says we value what engineers build with AI but not what writers build with AI. It says technical creation is legitimate collaboration and creative creation is fraud if you had help.

Here’s what I actually do.

I come to the table with a thought. Not a prompt. A thought. A memory, an experience, a connection I’ve made between two things that don’t obviously belong together. Something I noticed in an art store or on an offramp or in my grandmother’s kitchen that has been sitting in my chest waiting for a shape.

The thought exists before the AI does. It exists before the words do. It lives in a space that’s pre-verbal, pre-formatted, pre-anything that looks like a sentence. It’s a shape. A feeling. A topology. And I want to get that shape through the bottleneck of language and onto a page where someone else can see it.

That’s where the AI comes in. Not to think for me. To help me think ONTO THE PAGE.

I talk. I tell it the story. I tell it what I noticed and what it connects to and where the thread leads. And it helps me compile those thoughts into a structure. A frame. A template, essentially — not unlike the templates people use every day for resumes and business proposals and pitch decks without anyone accusing them of not really writing their own resume.

Then I go through it. I change things. I add my voice where it’s missing. I cut the parts that sound flat. There are always flat parts, because the AI doesn’t know what I know. It doesn’t know what the air felt like in the kitchen when I wrote about my Grandmother. It doesn’t know the weight of a tote that’s been sitting in the back of a car for months waiting for the boy without a home that I met one day and then never saw again. It doesn’t know the difference between a memory that matters and a detail that’s just filling space.

I know. That’s my job. The AI holds the frame. I do the painting.

The slop argument isn’t wrong. It’s just incomplete.

Yes, AI-generated content can be garbage. Vast quantities of it are. The internet is filling up with soulless, repetitive, algorithmically optimized nothing produced by people who have nothing to say and found a tool that says it faster.

But dismissing all AI-assisted creation as slop is like dismissing all food as fast food because McDonald’s exists. The tool isn’t the meal. The cook is the meal. The ingredients are the meal. The memory of your grandmother’s hands turning a bowl — that’s the meal. You can make it in a pristine kitchen or a campfire. The stove doesn’t determine the quality. The human does.

People were writing empty things long before ChatGPT. Ghostwriters have existed for centuries. Templates have existed for decades. Every politician’s speech is written by someone else. Every CEO’s memoir is assembled by a collaborator. Nobody clutches their pearls about those. But the moment the collaborator is artificial, suddenly the work is invalid. Suddenly the thought doesn’t count because the frame it’s sitting in was built by a machine. Suddenly the amount of em dashes in your article is evidence of a crime committed.

That’s not a critique of AI. That’s a bias against a tool you don’t understand.

Here’s what the slop critics are actually arguing about, whether they realize it or not: format.

They’re arguing about sentence structure. About word choice. About the way a paragraph is assembled. They’re looking at the arrangement of words on a page and asking “did a human hand place each one of these in this specific order?” Or “I can tell this is AI because of the em dashes.” Sir. Some of us were using em dashes before ChatGPT was a twinkle in Sam Altman’s eye. Some of us are just DRAMATIC PUNCTUATORS BY NATURE. Emily Dickinson used dashes like breathing and nobody accused her of being a language model.

“Too many em dashes.” “That sentence structure looks AI.” “The transitions are too smooth.” As if writing well is now suspicious. As if competence is evidence of fraud. As if the only way to prove you’re human is to write badly on purpose.

And I’m saying: who cares?

The thought is the thing. The story is the thing. The memory, the noticing, the connection between a glitter pen and a grandmother’s kitchen and the reason legacy matters — that’s the thing. That existed in my head before I opened any tool. That came from my life. My body. My years. My noticings. My grief and my joy and my tote full of sleeping bags in the back seat of my car.

If you want to argue about whether the AI arranged those words or I arranged those words, you’re arguing about the canvas. You’re standing in front of a painting asking what brand of brush was used. You’re reading a poem and asking what kind of pen wrote it.

The pen doesn’t matter. The hand does. And behind the hand, the eye. And behind the eye, the life.

The real test isn’t “is this AI-generated.” The real test is “is this yours.”

Can you trace every idea back to something you lived, saw, thought, or felt? Can you point to the sentence and say that came from a Tuesday in an art store when a woman asked me about acrylic ink and I recognized her courage because I’ve been her? Can you say this paragraph exists because my grandmother wrote a cheesecake recipe in glitter pen on a card with a duck on it and I didn’t realize until this morning that the glitter in my work comes from her?

If yes, the tool is irrelevant. The specificity is the proof. AI trained on the general population cannot produce YOUR grandmother. It cannot produce YOUR offramp. It cannot produce the particular weight of YOUR sleeping bags in YOUR car waiting for a boy whose name you never learned. That level of specificity is not in the training data. It’s in the human.

If no — if the work is generic, if it could have come from anyone, if there’s no fingerprint, no teeth marks, no scar tissue in the sentences — then that’s not an AI problem. That’s a human problem. And it existed long before any of these tools did. Slop is still slop, whether produced by human hands or AI. I suspect the real grudge has to do with the speed at which people who write with AI can produce work.

I have published over twenty books. I am building a mythology, a podcast, a body of interconnected work that spans fiction and nonfiction and memoir and humor and things that don’t have a genre because they crawled out of the same drawer. I use AI as a thought partner in much of this work. I use it the way a musician uses a studio — to capture, arrange, refine, and produce something that starts as a sound in my head and ends as a thing you can hold.

The sound was always mine. The studio just helped me record it.

If you want to dismiss that because of the tool I used, go ahead. But you’re not critiquing my work. You’re critiquing your own discomfort. You’re telling me more about your relationship with technology than you are about the quality of what I’ve made. You’re telling me more about your ability to see the thought behind the words. And you’re telling me exactly what kind of person you are, that dismisses rather than engages in the wondering.

So go ahead, point to the sentence that could have come from anyone. Let that be the thing you pick apart instead of the thought behind it.

The argument was never about AI.

It was about you gatekeeping how a thought gets on page.

Written with Claude. If that’s what you want to focus on, you’re proving my point.

Back to blog

Leave a comment

Your Name
Your Email