A prize-winning story published in Granta was (very likely) written by AI.

It’s another grim day at the human factory. There is strong evidence to suggest that a prize-winning short story published this week in celebrated literary magazine Granta was entirely generated by AI.

Jamir Nazir’s “The Serpent in the Grove,” was published in Granta in partnership with the Commonwealth Foundation Short Story Prize, which annually recognizes unpublished fiction from around the British Commonwealth. Nazir’s story, which follows a rum-drinking farmer who comes across an enchanted grove, was the recognized entry from the Caribbean region.

A set of judges chaired by award-winning novelist Louise Doughty appreciated the story’s “precise yet richly evocative” language, and selected the piece as a regional finalist from a whopping class of 7,806 entries across the board. But literary sleuths smelled a rat.

Ethan Mollick, a Wharton professor who studies AI’s effects on education and the workplace, broke down this “Turing Test of sorts” in a helpful Bluesky thread.

Off a hunch, Mollick ran the story through Pangram, a program that detects AI writing with 99% accuracy. (And because this is a slippery, evolving science, Mollick also cited independent research re: the tool’s efficacy.) Pangram has an extremely low false positive rate. And “The Serpent in the Grove” came back with 100% red flags.

The award-winning author has also proven hard to locate in meatspace. Nazir’s bio identifies him as “a Trinidadian writer of East Indian heritage whose work explores the cultural intersections of the Caribbean and the Indian diaspora.” Four days ago, The Jamaica Observer reported that he is 61. But there is remarkably little else to his digital footprint.

In 2018, Nazir appears to have self-published this book of inspirational poems, much-feted on Amazon. But he has no other published work that this reader could detect using the good old fashioned internet. And on his LinkedIn page, he is a frequent AI evangelist. “Generative AI won’t replace good leaders,” he claimed in a recent post. “It will expose poor ones.”

On close read, the winning story does have a few tells. Style-wise, Nazir favors parallelism, epistrophe, and lists of three [👀], all rhetorical devices that are associated with large language models. He reaches often for the poetic metaphor. Consider the passage below.

Wilfred’s rum-shop leaned into the road like a rotten tooth. Inside, boards blackened by smoke and sweat, the air sweet with cane and forgetting. Coins meant for rice or kerosene slid across the counter and came back white rum hot as apology. One drink opened the chest, two turned fear into courage’s cheap cousin, three steadied the hand enough to write the future in invisible ink.

Of course, style mileage should vary—and in this case, Nazir’s “melodic voice” is what endeared him to the judges. But one could still make the case that this graf has flaws a flesh and blood writer could be less likely to make. (Strained similes, etc.) We might spin the sting as damning, mostly, to Granta. As the human author Tony Tulathimutte pithed in an Instagram post yestereve, this news “should not disturb anyone who is at all familiar with what awards judges tend to favor.”

But it’s also just true that AI writing is getting harder to detect with the naked human eye. Anyway, the biggest bummer is to come. Winning a literary prize is one small step for LLMs, but it’s sure to be catnip for the pushers touting the technology’s creative potential.

My colleague James Folta called this news a “grim Rubicon.” And on X, the AI scholar Nabeel S. Qureshi called the kerfuffle a “major milestone” for AI evangelists (like Nazir). As The Hindu reported hours ago, the Commonwealth Foundation is currently reviewing its selection process as accusations mount.

If all goes off as planned, the winner of the Commonwealth Prize will be announced in June. In the meantime, you can find the finalist stories here, and judge for yourself.