by Christos Zabounis
Those of you who use – or, rather, those of us who use—artificial intelligence in our writing, the time has come to be afraid. An A.I. detector can determine whether a piece of writing has been produced, for example, by ChatGPT, Claude or Gemini. The latter, in particular, specialises in writing articles, letters, and even novels. Let us take a breath. Although Pangram, as it is called, is considered one of the most sophisticated tools of its kind, the results it provides are not 100% accurate. Its use, however, has created the biggest literary scandal in France in recent years. A group led by Samuel Fitoussi, a conservative-minded columnist for Le Figaro, exposed the extensive use of A.I. in the book by left-wing-supported writer Thelyson Orelien, using Pangram. C’était ça ou mourir (“It was either this or die”) was the first novel by the Canadian-Haitian author, dealing with the hardships of his migration. Le Monde, which submitted the book to the tool, estimated that more than 95% of it had been generated by A.I. Further tests raised the figure to 100% in several passages. When asked for its own opinion, the A.I. itself replied that Pangram’s results do not, in themselves, constitute proof. Alas! New investigations carried out on Orelien’s earlier writings revealed extensive plagiarism. An old trick, after all. In the latest development, the Goncourt Academy withdrew the book from the final shortlist for its prize, pointing to its mission of honouring literature written by human beings. What happens, however, when A.I. begins to write better than a human—as may perhaps be the case with Orelien, since those who read his book praised it highly? “Un grand roman, une voix” (“A great novel, a voice”), wrote the literary critic of Le Figaro before the scandal was exposed.
