Prix Pascal: If AI writes your book, are you still a writer?

There is now a new French literary prize for AI-assisted writing — the Prix Pascal de Littérature IA. The Pascal prize requires entrants to turn in an unpublished manuscript, fiction or non-fiction, in French, of approximately 3,00,000 characters, written with the help of large language models (LLMs), such as ChatGPT, Claude or Gemini. Writers must declare which AI model and prompts they have employed. The applications, to be submitted by November 30, consist of “a seed, a manuscript and a technical file, submitted simultaneously”. Ten finalists will be announced on December 4 and the winner will be revealed on December 15.The manifesto on the prize website reads:“This [submitted] work must be useful, or raise awareness, or help to solve in an original way one of the ecological, social and ethical challenges that threaten humanity. AI can now solve previously unsolvable mathematical problems. Here, it is asked to produce a text that will help humanity progress towards wisdom and the preservation of itself and its environment.”I love the way in which the manifesto strategically reframes tacit support for what goes against the very grain of the creative spirit as an essentially moral and ethical act. But of course! This is the way these things are usually done. The prize was founded by Philippe Rodriguez, Albin Serviant and Gilles Mentré. The first two are technology entrepreneurs, which too is no surprise. When creativity is outsourcedNotwithstanding their precarity, writers and artists have always been viewed with a combination of envy, fear and hatred. The imagination is a dangerous thing. It is ungovernable. It cannot easily be controlled or colonised. It does not seek permission. And that, to many, is frightening. What better way to colonise the imagination than to introduce AI into the writer’s world? Also, that way everybody — even the ones with little understanding of what a writer does — can acquire the overnight label of one. Where does this leave writers who have spent years honing their craft, writers who have survived by the skin of their teeth? The world needs writers more urgently than ever before, and more of us should let some form of creative practice into our lives.Two semesters ago, a few students from my Creative Writing course, at Shiv Nadar University, Chennai, confessed to what their worst fears were: that, having come of age in the post-AI era, they would find it a challenge to access and exercise their imagination — which, in their understanding, was foundational to the course. I was struck by their capacity for self-reflection, moved by what they recognised, quite rightly, as an imminent loss. This, too, is the other dark side of what we have got ourselves into: the loss of imagination is foundational not just to the stray creative writing course but to life itself.The argument that it is best not to be left behind, that AI is here to stay so writers must take the good things it has to offer and navigate the rest, is a specious one. The folks making this argument may be right about the inevitability of AI, but does that make any of it less disturbing or a tad more ethical? I don’t see how one can frame this as anything other than a pure black-and-white story. If you give AI an inch, it is very likely that you will take a mile from yourself.It is a slippery slope.You start by using AI as an instrument — perhaps to help you rewrite a sentence — and before you know it, you are likely to seek its help extensively to write the entire piece. Who then is the real author of the piece? After all, it is in the writing that writers do their thinking. I don’t think the answer lies in declaring AI use upfront. While this may address the ethical question to a small extent, it still does not change the fact that we are always going to ‘writing’ with training wheels on. The craft is in the writing and if you give that away what is left?There is another problem too: that of the asymmetry between the writer on whose work an LLM is trained and the user who then draws on that work to write in “collaboration with” that particular AI model. The original writer whose consent has very likely not been sought gets nothing from this entire process — a process which de-humanises creative work and repurposes it as training data.Can we tell who wrote that? Artificial intelligence illustrations are seen on a lapto with books in the background in this illustration photo on 18 July, 2023. (Photo by Jaap Arriens/NurPhoto via Getty Images) | Photo Credit: Getty Images As almost everyone from editors of literary journals on the brink of closure to writing teachers will tell you, it is often impossible to tell AI-assisted writing apart from human writing. In fact, you may find that some AI-assisted writing may be as good as if not better than writing by humans. It, of course, depends on the prompts used and how advanced the models are.The LLMs have been trained to mimic human syntax, and because humans frequently edit AI text or may even adopt AI-like phrasing, it is impossible to be absolutely certain which is which. Neither dedicated AI-detection tools nor human readers are able to establish clearly whether a piece of writing is AI-generated or authored by a human. Worse still, detection tools often throw up false positives. This is especially true when it comes to non-native English writing. There are also ways to bypass detectors. OpenAI, however, discontinued its own native AI detection tool in 2023 because of low accuracy rates.We are often told that AI-assisted text tends to exhibit certain stylistic patterns such as the em dash — which, incidentally, I use a lot — that you will find in it the overuse of certain words such as “delve” and parallel contrast structures such as “It is not just X, it is Y” or that it lacks “soul”. All these arguments were used in the controversy over Jamir Nazir’s 2026 Granta-Commonwealth Short Story Prize-winning entry “The Serpent in the Grove”.Last month, Éditions du Boréal, publisher of Haitian-Canadian writer Thélyson Orélien, declared that it had confirmed the author had used AI to write his novel C’était ça ou mourir (It Was Either That or Die). But in truth, there is no way to establish AI use beyond speculation.A lot of humans use em dashes and the word “delve”, and a lot of humans write badly and without a hint of this thing we call soul. And AI, one must remember, has been trained on both “good” and “bad” human writing. The argument that AI helps democratise literature or the creative practice doesn’t hold water either. Creative practice and literature should be made accessible to all. That will help democratise both. Outsourcing the work to a machine won’t.The joys of wrestling with wordsWhy is it important to reclaim human writing in all its glorious imperfection? I played a minor role in a project that looked at AI tools and how effective they are when it comes to literary translation. I set ChatGPT a prompt to translate a Tamil poem. I had translated the same poem a few days earlier. In a matter of seconds, ChatGPT threw up a perfectly decent translation of the poem. I had spent close to two hours coming up with the first draft.As I shut the laptop, I felt a great emptiness wash over me. I had been hearing repeatedly that AI tools would help writers and translators save and optimise time. But did I really want that saved time? Or did I want the joy, the sense of protracted, slow wrestling which is an inevitable part of the creative process?Writing has saved me over and over. The hours spent in its company were never wasted, not even when I have had no finished literary text to show for them. On the other hand, the quick literary product that AI generates leaves me feeling purposeless and bereft.The writer’s most recent book is Footnotes to the Mahabharata (2025).