A Treatise on Writing in the AI Age, Part 2
For most people, writing a blog post isn’t solitary work. It’s start with talking to your friend over coffee and the question they ask that changes your perspective and leads you to reframe the whole thing. Then there’s the spellchecker and grammarly that give you one-click fixes and improve your voice. If you’re lucky enough, you have an editor or a teammate or colleague willing to read it over and give you feedback that leads to your third or fourth or fifth revision. And none of them appears in your byline. You’ve never been asked to disclose a single one.
But now that AI has entered the arena, you’re expected to wear a badge.
The rule we never had
Ghostwriting is the clearest tell. For a century, public figures have published books they didn’t write. Not books they wrote with help. Books where someone else produced every sentence, and the named author’s contribution was a series of interviews and a final approval. The industry is respectable, professionalized, and almost entirely undisclosed. Readers buy the memoir, connect with the story, and nobody calls it fraud.
Let’s be clear about what that means. Our writing culture has already decided that full delegation of the actual writing is acceptable, as long as the ideas and the accountability belong to the named author.
So the standard was never “you must write every word yourself.” It couldn’t have been. Editors restructure. Translators rewrite entirely. Co-authors merge beyond untangling. Writing has always been more collaborative than the solitary-genius myth admits, and readers have always been fine with it.
The disclosure mandate invents a purity standard that never existed, then applies it to exactly one tool.
The label that can’t label
Even if we granted that AI is different in kind, the mandate would still fail on its own terms, because “AI-assisted” doesn’t describe anything.
Consider what fits under that one flag. Fixing typos. Brainstorming titles. Summarizing a paper you then read yourself. Arguing with a chatbot to stress-test your thesis. Generating an outline. Generating a draft you rewrite completely. Generating a draft you don’t.
Those are wildly different acts with wildly different implications for the reader. Some of them involve less delegation than hiring an editor. One of them is closer to ghostwriting. A single checkbox flattens all of it into one undifferentiated confession.
A label that carries no information isn’t disclosure. It’s ritual.
We’ve been here before
Every writing tool arrives to the same funeral music.
Photography was going to kill painting, and for decades photographers fought to be considered artists at all, because the machine did the work. Word processors were going to ruin prose, because revision would become too easy and writers would stop thinking before typing. Calculators were going to destroy mathematical minds. Spellcheck was going to breed illiterates.
Each time, the panic confused the tool with the thinking. Each time, we eventually figured out that the craft didn’t live in the mechanical layer. Painting survived because painting was never just image production. Writing survived the word processor because writing was never just typing.
I don’t say this to wave away every concern about AI. I recognize that AI can hollow out writing in a way spellcheck can’t, if you let it do your thinking instead of your typing. That distinction matters enormously. It’s the difference between a tool and a replacement.
But notice what that distinction is. It’s not about which tool touched the text. It’s about who did the thinking.
Which is exactly what a disclosure checkbox cannot capture.
What the double standard reveals
Why does one tool get a mandate when a century of ghostwriting got a shrug?
Because the mandate was never derived from a principle. If platforms had started from “readers deserve to know how work gets made,” they’d have built process transparency for everything, and the absurdity would have been obvious immediately. Imagine the disclosure form: Did you discuss ideas with your spouse? Did you read the competitor’s post first? Did an editor rewrite the conclusion?
Nobody wants that, because readers never needed it. What readers needed was a writer who stood behind the work.
There’s even a version of this buried in my own story, and nobody flagged it. I found it myself, reading the guidelines the moderation comment linked. My post went out under my personal account and talked about how one of the features we just launched helps to fill a gap teams face when it comes to writing skills. As far as I understand, that’s allowed on the platform. Company blogs share product content there every day. But the AI guidelines add a special rule. AI-assisted articles shouldn’t promote any business, program, or course, including your own. Trace the logic. A human-drafted post can discuss the product it’s about. The identical post with AI anywhere in the process cannot. Same words, same reader, different rules, and the only variable is the tool. That’s not reader protection. That’s a purity test.
The AI mandate exists because AI is new and frightening, and new fears demand visible responses. It’s not a standard. It’s a nervous gesture wearing a standard’s clothes.
What if we asked the question the mandate skips? Not “did a machine touch this text” but “did a mind own it.” That question has an answer worth knowing. The next essay is about what happens when we settle for the wrong question instead, because the cost isn’t hypothetical. It has a name, and the name is stigma.