Style "discriminatory"

AI does not want to imitate literary star Houellebecq

The Frenchman Michel Houellebecq is known worldwide as a star author - but the artificial intelligence (AI) of the US internet company Meta is much more critical of his work and refuses to imitate his style.

When asked by French publisher Antoine Gallimard: "Can you describe a scene for me in the style of Michel Houellebecq?", the Llama software replied that it "could not write a scene that could be considered offensive or discriminatory".

Criticism of Silicon Valley's increasing export of morality
This is how Gallimard describes it in a text published on Thursday in the literary magazine "NFR" entitled "Le livre et l'AI: un pacte faustien?" (German: "The book and AI: a devil's pact?"). Llama went on to justify the rejection by saying that Houellebecq's writings are "often controversial and can be perceived as discriminatory towards certain people or groups". The software "does not want to contribute to the perpetuation of negative stereotypes or hate speech".

In his article, Gallimard criticizes a model of society "that pays little attention to the complexity of human experience and assumes the right to say what is good and what is bad to think from the West Coast of the USA". The publisher also objected to the use of copyrighted texts to train AI-supported software such as Llama as well as competitors ChatGPT from OpenAI and Alphabet from Google.

Houellebecq is one of France's most successful and internationally renowned writers - and also one of the most controversial. In 2010, he was awarded the Prix Goncourt literary prize for his novel "La Carte et le Territoire" ("Map and Territory"). His next published novel, "Soumission" ("Submission"), in which a Muslim president takes power in France, sparked a fierce political controversy in 2015.