Jeanette Winterson asserts that “humans will always want to read what other humans have to say” (OpenAI’s metafictional short story about grief is beautiful and moving, 12 March). But she neglects the fact that every piece of writing, film‑making or art created by artificial intelligence and then consumed by humans is one less piece of writing, film-making or art created by a human being that will have the chance to be read and enjoyed by humans.
One can only wonder if Winterson would hold the same view were she now at the start of her career, rather than having 40 years of success behind her.
We are already seeing creative jobs like editing and graphic design being outsourced to AI in dozens of industries. Is Winterson really so naive to think that the same won’t happen in literature if AI writing continues to be championed by influential people like her and published prominently in outlets like the Guardian?
My question and challenge to the editor is simply this: when was the last time a short story written by a human being was published on your website in a position of such prominence? You have given a story written by AI a huge audience. Why not now do the same for an aspiring, young, human writer?
Henry Futcher
Norwich
It was refreshing to read Jeanette Winterson’s article about the creative potential of artificial intelligence. All literature expresses memory and is synthetic, drawing consciously or unconsciously upon human experience, culture and other literature. Neuroscience has shown that humans don’t “experience” experience as simply being present “in the moment” – it’s filtered by consciousness of earlier experience, expectation, memory and social context.
So, even the “I” in lyric poetry may be read as the most unreliable form of intimacy – their “feelings” may move us, while being as “inauthentic” and “trained” as an algorithm. Literature asks an intriguing and elusive question: “What would it be like to be someone else?” How fascinating to have that question posited by a machine. AI may only be as “other” as we are to each other. Its creative processes and outputs will surely help to illuminate our own.
Graham Mort
Emeritus professor of creative writing, Lancaster University
As a writer of fiction, I read with interest, and unease, the opinions of authors I admire on the creative writing skills of ChatGPT (A computer’s joke, on us’: writers respond to the short story written by AI, 14 March). In my day job supporting neurodivergent students with academic skills, I suggest the use of AI as assistive technology – rather than generative, to actually write their essays.
Recently, a student told me that he uses the “AI detector” GPTZero to ensure his written work does not register as AI-generated. I then tested this with a sample of my own work – the first paragraph of a novel. To my horror, it asserted that it was “moderately confident” that the text was “85% written by AI”.
What if agents and publishers use this tool and fail to question its “judgment”? Can I reassure myself that GPTZero assessed my work as it did because the paragraph was well written? Was this a result of the fact that, after years of extensive reading, I am my own large language model, at least in the realm of fiction? Times are hard enough as a writer without being plunged into an existential crisis of this sort, so I am clinging to that last idea as the most reassuring.
Cal Walters-Davies
Aberystwyth, Ceredigion