Concerns over the increasingly widespread use of A.I. tools to assist with writing continue to roil the literary world.
This week, the Commonwealth Short Story Prize drew attention from readers who suspected that one of its winning stories, “The Serpent in the Grove,” was written by A.I. Like all five regional winners of the competition, the story was published online by Granta, a British literary journal that has featured work by renowned authors like Kazuo Ishiguro, Zadie Smith and Salman Rushdie.
“We’ve taken stock of the comments and tried to be very systematic in our understanding of some of the perspectives and tried to look at ourselves internally to see if we feel that our process to date has been robust enough,” said Razmi Farook, the director-general of the Commonwealth Foundation, which administers the prize, in a video call with The New York Times. “We’re confident in the rigor of our process, but we’re conscious that this is an evolving technological environment.”
Granta, which did not respond to questions sent by The Times before publication, took a different stance in its multiple public statements.
“We showed Claude.ai the story and asked whether it was A.I.-generated,” Sigrid Rausing, the publisher of Granta, said in a statement. “The response was long, concluding that it was ‘almost certainly not produced unaided by a human.’”
Rausing added: “It may be that the judges have now awarded a prize to an instance of A.I. plagiarism — we don’t yet know, and perhaps we never will know.”
On its website, the magazine posted a statement above each of the five winning stories distancing its editors from the contest’s selection process.
“Granta editors were not involved with these stories or their selection beyond copy-editing them upon receipt,” the statement reads in part. “The suggestion that writers have submitted material not authentically their own is a charge we take seriously, but until definite evidence comes to light we will keep these stories on our website.”
The story’s author, Jamir Nazir, is described in an accompanying bio on Granta’s website as “a prolific poet and author, with books published and others forthcoming.” Among the little other work that Nazir appears to have published is a self-published poetry collection, “Night Moon Love: Poems for All Who Have Loved or Dreamed of Love.” Nazir did not respond to a request for comment.
This year’s competition received around 7,800 entries. Of the five chosen as regional winners, one will be named the overall winner on June 30.
The speculation surrounding “The Serpent in the Grove” came on the heels of news, reported by The Times in March, that Hachette Book Group was halting plans to publish “Shy Girl,” a horror novel, in the United States, and discontinuing sales of the book in the United Kingdom. The move followed widespread allegations online that its author, Mia Ballard, had relied heavily on A.I. to write the book. (Ballard told The Times that an acquaintance she hired to edit a self-published version of “Shy Girl” had used A.I., but that she herself did not.)
Earlier this week, The Times also found that the author Steven Rosenbaum had included a number of fictitious or misattributed quotes generated by A.I. in his recent nonfiction book about artificial intelligence, “The Future of Truth.” (In a statement, Rosenbaum said that the book had “a handful of improperly attributed or synthetic quotes” and that he had started his own investigation.)
Authors have taken varying public stances on the role of A.I. in their creative processes.
The author Coral Hart, for example, has been open about using A.I. to help her write; last year, she self-published more than 200 romance novels, with the help of Anthropic’s Claude.
The Nobel Prize-winning Polish author Olga Tokarczuk caused a minor uproar in the literary community this week after telling an audience in Poland last week that she used A.I. while writing her latest novel. Tokarczuk responded with a statement shared by her publisher.
“I make use of artificial intelligence on the same principles as most people in the world — I treat it as a tool that allows faster documenting and checking of facts,” Tokarczuk said. “None of my texts, including the novel that will appear in Polish this fall, has been written with the help of artificial intelligence — except for using it as a tool for faster preliminary research.”
For publishers and prize juries, A.I. presents a challenge with no clear solutions. The accusations made against “The Serpent in the Grove” and other works have revealed how tricky it can be to determine whether a work was written by a human.
In the days following the story’s publication, readers on social media and online forums pointed out signs of A.I. writing tics, such as excessive use of metaphors and similes; nonsensical figurative expressions (one line in particular, “she had the kind of walking that made benches become men,” drew a lot of flak); and negative parallelism, which is more commonly known as the “not X, but Y” construction.
But these traits often appear in human writing as well. To confirm their suspicions, some readers, including the Wharton School professor and A.I. researcher Ethan Mollick, ran the story through A.I. detectors like Pangram, which suggested that the text was 100 percent A.I.-generated.
Still, some experts caution that these tools are not entirely reliable.
“The A.I. text detectors make lots of mistakes, particularly on creative writing that uses unusual constructions that may not be typical in what Pangram has trained their system on,” said Nicholas Andrews, a senior research scientist in computer science at Johns Hopkins University whose work focuses on A.I. and machine learning.
Jack Grieve, a professor of corpus linguistics at the University of Birmingham, pointed out that without controlling for variations in dialect, topic, genre and prompts, relying on A.I.-detection tools can be risky. To his eye, he wrote in an email to The Times, the short story was neither obviously A.I.-generated nor obviously not A.I.-generated.
“The technology isn’t quite there, and because of its imperfections, it leads people to think things are black and white when they’re not necessarily so,” Farook said. She added that for organizations, it is important to avoid having “a knee-jerk reaction to the general hysteria around these issues at the moment.”
This is especially the case, Farook added, because of the risks to writers that false accusations carry.
“These are people with personal stories, writers facing their own personal challenges, and writing is a very vulnerable tool,” Farook said. “The prize brings those people to the fore. They’re having to contend with something new. They’re often unpublished writers, so there’s a duty of care and consideration we need to take into account.”
The post Was a Story That Just Won a Literary Prize A.I.-Generated? appeared first on New York Times.
