Our words are worth more than their pictures. / Literature, photography, and artificial intelligence

Recently, image generators have been able to produce images from "prompts," that is, a sentence of text that an artificial intelligence interprets from its database to produce a result.

I was curious to compare these software programs with literary extracts describing places or characters.
The results are both surprising and expected.
On the one hand, any text that is even remotely literary is completely beyond their comprehension: Proust, Perec, etc. It's all Greek to these programs, which interpret them in a totally incoherent way: Simone de Beauvoir's family portrait depicts the same child eleven times! On the other hand, the programmed desire to create "beautiful" images, whatever the subject, often results in complete misinterpretations.
Otherwise, the results are fairly consistent in their "cheap" aesthetic. I honestly tried to edit these photographs with the same care as my own.

I undertook this work as an experiment and a basis for reflecting on what we have the right to expect from a photographic image. The answers are obviously multiple.
If it's a matter of creating a background image for a laundry detergent advertisement, whether you start with images copied from the internet or invent them from scratch, it doesn't matter. But if it's a matter of so-called original works, verified origin must be the rule, as with any work.

Ethical questions remain entirely open, and media literacy is becoming more important than ever to integrate into all school curricula.