It has become the most frequently asked question in modern writing: “Have you used AI?” It’s also the wrong question. A writer who asked a chatbot to revise a sentence, one who asked for careful fact-checking with sources, and one who suggested a few ideas and said, “Write me a comment!” represent completely different uses for this new type of support. I would argue that most authors should use AI in some form. What readers need is a better question and a scale to help answer it.
Last month, billionaire investor Stanley Druckenmiller admitted to using AI to write a letter Wall Street Journal op ed attack Finance Minister Scott Bessent’s interventions in the bond market. Readers quickly recognized the telltale prose (“It wasn’t liquidity management, it was price management”) and the AI detector Pangram dotted the piece 100 percent AI
When NOTUS asked him about itDruckenmiller was not ashamed. “Of course I used AI,” he said. “I’m not embarrassed… I write everything with AI now, for the same reason I use a calculator when solving math problems.”
The magazine was similarly unapologetic. “AI is a fact of modern life,” said editor Paul Gigot said in a statement. The next day, in one SplitGigot went further: “Mr. Druckenmiller’s sin against the media Pharisees was admitting their use.”
Druckenmiller’s flawed analogy aside (math has only one right answer, writing doesn’t), the saga revealed the real problem with writing in the AI age: We keep asking writers a yes-or-no question (have you used AI?) when the answer increasingly lies somewhere in between. Druckenmiller said he rejected many of the chatbot’s suggestions and that the AI didn’t write “the whole thing.” Pangram, on the other hand, achieved 100 percent AI. When it comes to writing and reporting, nuance matters. Readers want to know whose idea it was and whose words they are reading: a person or a chatbot.
I have been working with AI for years. At NewsGuard, a company that rates the reliability of news sites, I tracked AI-generated misinformation, namely The Wall Street Journal I created my own AI propaganda page for $105 to show how easy it is. Today I’m running Newsreela media startup where we vet independent journalists before they can publish. Despite this review, readers still want to know how much AI went into what they read.
So, here’s my suggestion: a simple, self-reported label, printed with the byline or at the end of the article. Something like this:
The “Who wrote that?” scale
- H5 human: No AI was used at all. (Spell check doesn’t count.)
- H4 supported: AI helped with everything but writing: research, fact-checking prompts, headline ideas, grammar corrections. No sentence has ever been rewritten, and the prose is entirely the author’s.
- H3 polished: The author wrote the full draft, then the AI rewrote parts of it (streamlined sentences, shortened paragraphs, rearranged sections).
- H2 co-author: The author and the AI went back and forth. Some paragraphs started with the model, others with the author, but the argument came from the author and the author made a real correction.
- H1 Ghostwriting: AI wrote the first full draft from the author’s own notes, interviews, or reports. The ideas come from the author; most sentences are not.
- H0 AI Slope: AI wrote it from a command prompt alone, with minor changes at most. The ideas, structure and voice come from the model.
To be clear: I’m not the first to suggest this. The open source AI Usage Scale suggests six very similar tiers and at least one publisher, The Datatech Times outputs an AI level next to each byline.
However, the introduction of such scales is rare. That shouldn’t be the case. Writers are already fearful of an AI witch hunt: some are cutting the hyphens from their prose, and The is coming out in April magazineis James Taranto compared Pangram becomes a “defamation machine” that fuels false accusations.
A self-reported scale such as “Who wrote that?” could protect authors who use AI for research, editing or brainstorming. Once you reveal how you wrote something, there is nothing left to reveal.
Readers care less about an author using AI than about being misled. For example, voters are not upset about a president using a speechwriter because everyone knows that this has long been a common practice in politics, great speakers like Abraham Lincoln notwithstanding. The rich and famous have regularly hired ghostwriters, speechwriters and public relations people to develop a language known by their name. .
Of course, since “Who Wrote That?” Although the scale is based on self-reporting, some authors will be tempted to obscure the truth. But a label gives readers an answer without having to read an article through Pangram, and knowing that detectors exist should keep most writers honest. After all, claiming a story you wrote is H4 and getting caught at H0 is much more embarrassing than just saying it’s H1.
In 1845, a journalist accused Alexandre Dumas, author of The Three Musketeers and The Count of Monte Cristo, of running a “novel factory” staffed by hidden employees. Dumas sued and won. But the core question of this fight was never resolved: Who wrote this? If we can’t stop people from asking this question, we can at least start answering it honestly. Oh, and this article I’m sharing is H3 polished: The author wrote the full draft, then the AI rewrote parts of it (streamlined sentences, shortened paragraphs, rearranged sections).
The opinions expressed in Fortune.com comments are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Assets.