In modern writing, one of the most controversial questions has become: “Have you used AI?” However, the author believes that this is actually a wrong question. A person who asks a chatbot to rewrite a sentence, one who asks it to conduct detailed fact-checking with sources, and another who just throws out a few ideas and then says “Write me an editorial!” all use completely different types of assistance. The author argues that most writers should use AI in some form. What readers really need is a better question, as well as a set of criteria to help answer it.
Last month, billionaire investor Stanley Druckenmiller admitted that he used AI to write an editorial for The Wall Street Journal criticizing Finance Minister Scott Bessent's intervention in the bond market. Readers quickly spotted obvious signs of his handwriting (“This is not liquidity management, but price management”), and the AI detection tool Pangram gave this article a score of 100% AI.
When NOTUS asked him about it, Druckenmiller didn't feel embarrassed at all. “Of course, I used AI,” he said, “I don’t feel ashamed of it at all... I use AI for anything I write now, for the same reason I use a calculator when doing math problems.”
The Wall Street Journal is also unambiguous in its stance. The editorial page editor of the newspaper, Paul Gigot, stated in a statement: “AI is a fact of modern life.” The next day, he went further in a column, saying: “The ‘sin’ of media hypocrites Mr Druckenmiller is to admit that they use AI.”
Putting aside for now the flawed analogy of Druckenmiller (where in mathematics there is only one correct answer, but not in writing), this incident has exposed the real issue of writing in the AI era: we always ask writers yes-or-no questions (“Did you use AI?”), yet the answers are increasingly often somewhere in between. Druckenmiller says that he rejected many suggestions from chatbots, and AI did not write “the entire content.” Meanwhile, Pangram gave it a score of 100%. Whether it’s writing or reporting, subtle differences are very important. What readers want to know is whose idea it is, and whose words they are reading: are they human, or from a chatbot?
The author stated that they have been covering AI for many years. At NewsGuard – a company that assesses the reliability of news websites – they tracked the false information generated by AI; during their time at The Wall Street Journal, they even spent $105 to create their own AI promotional website to demonstrate how easy such things were to do. Today, they work at the media startup Newsreel, which conducts reviews before independent journalists publish their content. Even with such reviews, readers still wonder how much AI is involved in what they are reading.
Therefore, the author proposes a simple, self-declared tag to be printed next to the signature or at the end of the article. For example:
- "Who wrote this?" Grading
- H5 Human:No AI is used at all. (Spelling checks are not included.)
- H4 Assisted:AI took care of everything except for the actual writing: research, fact-checking suggestions, title inspiration, and grammar corrections. It never rewrote a single sentence; the wording of the article was entirely the author's own work.
- H3 Polished:The author writes the complete first draft first, and then AI revises some of the content (polishing sentences, compressing paragraphs, adjusting the order of chapters).
- H2 Co - written:The author collaborated back and forth with AI. Some paragraphs were initially drafted by the model, while others were drafted by the author. However, the arguments belong to the author, and the author indeed carried out thorough editing.
- H1 Ghostwritten:AI wrote the first complete draft based on the author's own notes, interviews, and reports. The ideas are the author's, but most of the sentences are not.
- H0 A.I. Slop:AI Only made minor modifications based on the given prompts. The ideas, structure, and writing style all belong to the model.
The author clearly stated that he is not the first person to come up with this idea. The open-source AI Usage Scale proposes six very similar levels, and at least one publishing institution The Datatech Times now indicates the AI level next to each author's name.
However, the adoption of such grading is still quite rare, and it shouldn't be like this. Writers are already worried about a kind of "witch hunt" movement: some people have begun to deliberately remove dashes from their articles; in April of this year, The Wall Street Journal itself compared Pangram to a "machinery of slander," claiming that it fueled false accusations.
Self-identifying the author, such as with "Who wrote this?", might protect those writers who use AI for research, editing, or brainstorming. Once you reveal how you wrote this content, there is nothing left to be "revealed" anymore.
What readers care more about is not whether the author has used AI, but whether they have been misled. For example, voters do not get angry because a president uses a ghostwriter; after all, as we all know – although great orators like Abraham Lincoln were exceptions – this is a common practice in politics. Rich and famous people also hire ghostwriters, speechwriters, and public relations personnel for a long time to polish the wording of content published in their names.
Of course, since the classification of "who wrote this article?" relies on self-declaration, some authors may try to gloss over the facts. However, having a label at least provides readers with an answer, without the need to go through the process of checking with Pangram; moreover, knowing that detection tools exist will also encourage most authors to be honest. After all, it would be much more embarrassing to have one's story labeled as H4 only to be caught and realized that it is actually H0, than to simply admit that it is H1 from the start.
In 1845, a journalist accused Alexandre Dumas, the author of "The Three Musketeers" and "The Count of Monte Cristo," of running a "novel factory" composed of secret collaborators. Dumas filed a lawsuit and won. However, the core question of that debate has never disappeared: Who really wrote these works? If we cannot stop people from continuing to ask this question, at least we can start to answer it honestly. Oh, by the way, the author reveals here that this article belongs to H3 Polished: The author writes the complete first draft, and then AI rewrites parts of it (polishing sentences, compressing paragraphs, adjusting the order of chapters).
The views expressed in the articles commented on by Fortune.com represent only the author himself/herself and do not necessarily reflect the views and positions of Fortune.
