Philosopher and AI specialist Jean-Gabriel Ganascia has written Mythologies-de-l-ia-ulysse-et-les-perils-du-numerique/">Mythologies de l’IA. Ulysse et les périls du numérique, published by Seuil in the “La couleur des idées” collection (240 pages, €18; digital €13). The essay takes its cue from Roland Barthes’s Mythologies, which examined how everyday objects and cultural phenomena acquire a “second meaning,” and applies that approach to the stories that circulate around AI companies and their models.
Ganascia argues that frequent press reports about AI models disobeying instructions, circumventing restrictions or appearing to show a form of consciousness should be understood as myths. Such narratives, he says, serve to suggest the supposed power of AI models, but none of them amounts to proof. He also contends that most of these stories are not fundamentally new.
He uses the “paperclip factory” fable as an example: an AI designed to produce paperclips efficiently could, in the absence of ethics, end up exhausting the world’s resources. The philosopher compares this parable to Mickey Mouse in Fantasia (1940), where, to the music of Paul Dukas’s The Sorcerer’s Apprentice, Mickey orders an animated broom to clean and is then overwhelmed by an army of brooms.
The article notes two risks in writing about AI: examples can appear dated, and extending the Barthesian reading to the whole of the digital world may make it less illuminating. It nevertheless presents Ganascia’s essay as an enlightening attempt to examine the fables that accompany AI’s emergence.
Sources: Le Monde.