The key facts
- Lukyanenko considers texts written with the help of neural networks 'not literature' and opposes awarding prizes to such authors
- Thelisson Aurelien's novel 'It Was This or Die' was removed from the longlist of the Goncourt Prize for the use of AI
- The writer and the publisher reject the accusations based on the Pangram detector
Lukyanenko's statement
Science fiction writer Sergei Lukyanenko told RIA Novosti, that authors who create books with the help of neural networks should not receive literary prizes. 'If some author of ours declares that he wrote a book with the help of neural networks, then let the neural networks be awarded the prize. In any case, I will be strongly against such authors receiving anything in literature. This is no longer literature. In any case — not for humans,' he said. In his view, writing books should remain a creative process for the author and bring pleasure.
Context: the Aurelien case
The statement came amid a high-profile scandal in France. In September, the novel by Canadian writer of Haitian origin Thelisson Aurelien, 'It Was This or Die' (C'était ça ou mourir) was removed from the longlist of the Goncourt Prize for the use of artificial intelligence. The trigger was a message from an anonymous user under the nickname Balance ton Claude on the social network X: according to the American AI detection program Pangram, the novel was almost entirely written by a neural network. The publisher Les Éditions du Boréal came to the defense of the writer, who is called a literary genius in France, and Aurelien himself rejected all the accusations.
What remains unclear
Lukyanenko's position is the personal opinion of a writer, not an official rule of any prize: how exactly literary juries will verify the use of AI does not follow from the publications. In the Aurelien case, the conclusions of the Pangram program are disputed by the author himself and the publisher, and there is no independent confirmation in the sources that the novel was written by AI. Thus, the question of the criteria for the admissibility of neural networks in literature and the reliability of detectors remains open.

