Where Assistance Ends and Authorship Begins

The real question about AI in writing is not whether it was used, but whether the human remained the main contributor.

Sign up for my Substack daily AI newsletter here.

Article link: https://johanosteyn.substack.com/p/where-assistance-ends-and-authorship?r=73gqa&utm_campaign=post&utm_medium=web&showWelcomeOnShare=true

Follow me on LinkedIn: https://www.linkedin.com/in/johanosteyn/

Two writers both use AI. Luke Stoffel is an American author with dyslexia. He used AI to fix weaknesses in his writing and to plan his marketing, and in May his memoir won an award from the Independent Book Publishers Association. Coral Hart is a South African author who uses AI to release more than two hundred novels a year under several pseudonyms. She says the business is profitable, though none of the books has become a bestseller. Few readers would say these two did the same thing, and that difference is where the debate about authorship really lies.

CONTEXT AND BACKGROUND

The debate has turned sharp. The writer Thelyson Orelien was dropped from consideration for France’s Goncourt Prize after an anonymous account on X claimed, citing the AI detector Pangram, that his novel was written almost entirely by artificial intelligence. He denies using AI.

Publishers are responding with labels. Books By People allows limited AI use for spelling and grammar. The Authors Guild’s self-certified mark now covers about six thousand five hundred titles. Writers already use these tools widely. A University of Cambridge study found that a third of UK novelists use AI, mostly for non-creative tasks such as searching for information.

INSIGHT AND ANALYSIS

Asking whether a writer used AI is the wrong question. Sebastian Guggolz, who heads the German Publishers and Booksellers Association, put it plainly at the Frankfurt Book Fair: “There’s no simple ‘with AI’ or ‘without AI.’” Writing with AI sits on a spectrum. It runs from spelling and search, through suggestions on phrasing, to drafting whole chapters.

A better test is whether the human remains the net contributor. A writer who conceives the story, shapes its structure, chooses its words and revises it is still its author. When the machine supplies the ideas, the structure or the sentences that make the work what it is, the name on the cover describes a manager rather than a writer.

I apply the same test to my own work. I publish an article on artificial intelligence every day, and I use AI to help me plan each one, structure the argument and find references. I then check every reference myself and research it thoroughly, which is a great learning experience for me. Before anything is published, I make sure the article reflects my view and says what I would have said. I believe this is a good use of AI, provided it is used responsibly, and the thinking remains mine. The tool speeds up the work, while the judgement and the accountability stay with me.

I have previously written about this in the context of AI detectors. They estimate statistical resemblance, which reflects style rather than who did the thinking, so publishers should stop treating their scores as evidence. The Authors Guild makes a similar point. It warns that publishers should never cancel a contract or pull a book on accusations or detector results alone.

IMPLICATIONS

Publishers need graded disclosure rather than a yes-or-no question. Contracts should ask what AI did and at which stage, and set out a fair process before any cancellation. Writers should keep their notes and drafts, which are a better defence than any detector score.

Business leaders should apply the same test to reports, speeches, board papers and strategy documents: the named person must own every word. The Stoffel case also warns against blunt bans, because assistance widens who gets to write.

CLOSING TAKEAWAY

Authorship was never about purity. Writers have always relied on editors, dictionaries and the books that shaped them, and what made them authors was that the work was theirs. AI does not change that test. It makes the test harder to apply and more important to state. The line falls where the human stops being the main contributor. A name on a cover is a promise that a person stands behind every word.

Author Bio

Johan Steyn is a prominent AI thought leader, speaker, and author with a deep understanding of artificial intelligence’s impact on business and society. He is passionate about ethical AI development and its role in shaping a better future. Find out more about Johan’s work at https://www.aiforbusiness.net