Where Does the Author End and AI Begin? Drawing a Creative Boundary

Where Does the Author End and AI Begin? Drawing a Creative Boundary

A writer can use artificial intelligence to organise notes, challenge an outline or find a better way to explain an idea. The same writer can also ask a model to generate an entire chapter. Both activities may be described as ‘using AI’, yet they are not remotely the same creative decision. That gap is becoming one of the most important questions in modern authorship. If we want readers to trust what we publish, we need to explain not only whether AI was involved but what it actually contributed.

The problem with a yes-or-no question

Ask an author whether they use AI and you may get a defensive answer before the conversation has even begun. Some writers hear the question as an accusation. Others take it as an invitation to demonstrate how technologically adventurous they are. Neither response tells us very much about the finished book. A simple yes or no hides the difference between research support, administrative help, substantive rewriting and generated narrative.

The better question is: which decisions remained with the author? Who invented the characters, chose the central conflict, shaped the plot, determined the voice and revised the final prose? If the answer is a human writer, with software assisting around the edges, that is a different creative process from one in which the machine supplies most of the narrative and a person selects what to keep.

A useful creative continuum

I find it helpful to think in terms of a continuum rather than two opposing camps. At one end sits wholly human-led creation, using ordinary tools such as dictionaries, search engines and word processors. Moving along the line, AI might help organise research, identify continuity problems or suggest alternative structures. Further along, it may substantially rewrite paragraphs, produce dialogue, develop scenes or generate a first draft. At the far end, a person may supply little more than a prompt and publish the resulting text.

These positions should not be treated as morally identical or automatically condemned. The important point is that each carries a different responsibility. Research assistance still requires fact-checking. Structural suggestions require the author to make the artistic decisions. Substantive generated passages raise questions about originality, quality, rights, contractual promises and disclosure. A mostly generated book should not be marketed in a way that misleadingly implies the prose was entirely written by a human.

What remains uniquely mine

As a science-fiction author, I spend a great deal of time imagining technologies that do not yet exist. It would be strange to dismiss a real emerging technology without exploring it. I use AI where it helps me think, organise or communicate, particularly around research, visualisation and the business of bringing stories to readers. But the central ideas, plot, characters and substantive narrative are my responsibility. Those are not administrative tasks I want to outsource.

A character's choice has meaning because of the history I have built around that character. A twist works because of what I chose to reveal and conceal. A scene matters because I decided what the reader should feel and why. AI can offer an alternative or expose a weakness in my plan, but accepting or rejecting that suggestion is part of my craft. The responsibility for the finished story cannot be delegated to a tool.

The invisible work behind a finished page

Readers usually see the polished result, not the dozens of discarded openings, revised timelines and uncomfortable decisions that produced it. That invisible work is precisely why the creative boundary matters. An author does more than assemble fluent sentences. We decide what is worth saying, what should be left unsaid and what kind of emotional or intellectual experience a story should create.

There is also a practical risk in allowing AI-generated prose to dominate a manuscript. Fluent language can disguise shallow motivation, repetitive rhythm or contradictions that are difficult to notice when reading quickly. A model may produce an apparently convincing answer without understanding the larger artistic intention. An author who relies on such output still has to take responsibility for every weakness that reaches the reader.

Transparency without turning every book into a disclaimer

I do not believe readers need an inventory of every spelling suggestion or research query. They do, however, deserve accurate information when AI has made a material contribution to the work they are buying. The appropriate level of disclosure may depend on the publisher, retailer, contract, genre and nature of the contribution. There is no single sentence that resolves every circumstance.

A practical policy should distinguish between tools that support a human creative process and tools that generate substantive content. It should also cover synthetic narration, illustrations and marketing assets, where licensing and audience expectations may differ from those for the written text. Whatever the policy, it must be truthful and applied consistently. The point is to build confidence, not to create another badge for people to argue over.

Four questions to ask before publishing

First: can I explain the creative contribution I personally made? Keep drafts, notes and revisions that demonstrate how the work developed. Second: have I checked factual claims and references introduced through AI-assisted research? A convincing-looking citation is not evidence that a source exists or says what the model claims. Third: do I have the necessary rights to any generated imagery, narration or other material? Review service terms and publishing requirements rather than assuming all output is unrestricted.

Fourth: would a reasonable reader or contractual partner feel misled if they knew exactly how this material was created? That is a useful test because it focuses on the substance of the contribution rather than a fashionable label. If the answer is yes, the presentation or disclosure probably needs work. If the answer is no, explain the process plainly and move on.

The next stage of authorship

The most productive future for authors is not one in which we pretend AI has no value, nor one in which creative responsibility becomes an optional extra. It is a future where tools expand what a writer can investigate and achieve while the writer remains accountable for the work. That requires judgment, transparency and a willingness to define boundaries before the market defines them for us.

I want to keep exploring the possibilities of AI without losing the reason I became an author: to tell stories that begin in my imagination and mean something to me. The technology can be part of the workshop. It should not quietly become the author of the workbench's most important creation. Where would you draw your own line between useful assistance and giving away the story?

Further reading and author resources

The Authors Guild — resources on AI, copyright and authors' rights: https://authorsguild.org/

UK Society of Authors — advice and campaigning on AI and creators' rights: https://societyofauthors.org/

UK Intellectual Property Office — copyright guidance: https://www.gov.uk/government/organisations/intellectual-property-office

About the author

Rob Frankson writes original science fiction and explores the relationship between technology, creativity and storytelling. Learn more: https://www.121minutes.com/about

Stay connected

For more articles about AI, writing and the creative life, subscribe to 121minutes: https://www.121minutes.com/subscribe

AI & Editorial Transparency

AI & The Author is edited and published by Rob Frankson. Artificial intelligence is used to assist with news research, initial drafting, content organisation and supporting imagery. All articles are reviewed and, where necessary, edited by Rob Frankson before publication. The opinions, editorial position and final decision to publish remain the author's.