How long before AI writes a Bestseller? A Literary Thought Experiment

How long before AI writes a New York Times bestseller?

It’s a question that lingers like a subplot, unresolved, faintly unsettling, impossible to ignore. Earlier this year, a publishing data analyst sparked headlines by predicting that an AI-written book could top bestseller charts by 2030. 

The clarification followed quickly: simple books. Joke books. Puzzle books. Nothing too emotionally charged. The literary world exhaled. Slightly.

But the idea lingers. So, how long before an AI writes a story that truly moves us? A novel as good as any that a best-selling author could write.

We’re already living in an age of algorithmic fluency. AI can mimic tone, complete sentences, even riff on genre tropes with eerie accuracy. It can suggest plot points, edit grammar, and help writers brainstorm. The tech is getting faster, more sophisticated, and less obviously robotic.

It is also increasingly, on the surface at least, hard to identify as having been written by AI—and, no matter what anyone says, the em dash is not conclusive proof.

Yet if you delve deeper and read more closely when AI-generated fiction is tested for quality, one pattern persists: the prose is often smooth, but sterile. Characters behave predictably. Emotions are reported, not felt. Narrative arcs emerge with the tidy rhythm of a formula rather than the grit of lived experience.

And that, it seems, is where the wires show.

Beyond the formula: Why human stories matter

Storytelling, at its best, is messy. It’s intuitive. It contradicts itself. A human voice carries memory, rhythm, cultural tension and emotional risk. It stumbles. It leaps. It occasionally fails, and that failure can be more interesting than perfection. The writing does things, and goes places, you cannot predict.

AI, by contrast, has no past. It borrows from everything and originates nothing. Its stories are collages, extrapolated from vast datasets of already-written work. Is it clever? Uncannily. Is it creative? Arguably not.

When you read a novel that stops you cold, when a sentence feels like someone wrote it with your own breath, the question becomes, can an algorithm know what that’s like?

Writers as craftspeople, not just creators

There’s another angle here. For many writers, creativity isn’t some mysterious gift; it’s labour. It’s revision, stubbornness, structural doubt, and starting over. AI might be able to mimic the surface of that effort, but not the soul of it. And writing does have soul. It’s that undeniable quality that lifts a really good novel.

This isn’t to say that AI has no place. Some writers are already using tools to develop ideas or unblock stalled scenes. It can be a kind of scaffold, a neutral sounding board. But few would hand over the reins.

Although there are some who are going further. People looking to cut corners to see if they can be writers, too. They are trying to get AI to do the heavy lifting and pretty much everything else. 

The AI books are out there 

No one knows for sure if these books are out there, but my take is they are. 

And it’s not just speculation. There have already been a few widely reported cases where writers have been publicly caught using AI to generate significant portions of their work—and paid the price.

In one high‑profile incident, two “romantasy” authors—K. C. Crowne and Lena McDonald—were called out after readers discovered lazy AI prompts left behind and embedded within the text of their published novels. Ouch. 

Passages included notes like “rewrite to match J. Bree’s tension-filled style,” clearly the output of a chatbot rather than a human editor. McDonald later admitted she used AI to edit and “shape parts” of her book and defended herself by saying her goal was always to entertain, not mislead.

Similarly, in the case of the novella Death of an Author by Stephen Marche (under the pen name Aidan Marchine), roughly 95 per cent of the text was generated by AI tools such as ChatGPT and Cohere, including cover art and blurbs. While the experiment prompted conversation about AI’s creative potential, many critics found the resulting narrative emotionally flat and mechanically structured.

Then there’s the broader ecosystem of self‑published works on Amazon, where lazy opportunists have churned out scam titles, in education, biography or genre fiction, prompted by AI models, often scraping and lightly rephrasing material from real authors without attribution. 

The Authors Guild has documented several cases of "sham books", including fake biographies of real writers, thinly disguised rehashes of memoirs and scam guidebooks—all self‑published, often very cheaply, and misleading readers in the name of profit.

These episodes tend to follow a familiar arc. First, reader suspicion: a flat voice, strange phrasing, unexpected plagiarism—or, in extreme cases, entire AI‑style instructions left unedited in the manuscript. Then, public backlash. Then, defensive statements and sometimes retractions or cancellations. But the real risk, the slow creeping normalisation of AI-written fiction, begins before any headline. When autopilot storytelling becomes routine, the line between authored and automated blurs until it barely exists.

So yes, those AI books are likely already out there. Not always labelled, not always easy to detect, but quietly making their way into the literary marketplace.

For now, at least, literature, the real work, remains beautifully human. Mostly. 

The Final Chapter…?

Will AI eventually write a bestseller? Yes. If we define “bestseller” purely by sales or formulaic appeal, especially in niche or genre-based formats.

But if we define it by something else. An emotional connection, originality, the ache of real human insight, then the answer might not be no, but not yet.

Maybe not ever.