-Analysis-
PARIS — How can we continue to trust a text, an image, or a video in the age of artificial intelligence? The question of trust in information has been around for a long time, as we know, but the emergence of powerful tools such as Chat-GPT for text, or Midjourney for photos, and many others, transforms the question into a potential nightmare.
The European Union’s executive body, the European Commission, wasted no time in raising the question of how to regulate these technologies, which risk transforming the information space into a jungle. Yesterday, even before the major European law that is being prepared – the AI Law – the Commission took the lead.
It calls on digital platforms to define an AI label that will enable users to know whether a text, photo or video has been generated, in whole or in part, by artificial intelligence. The aim is to limit the explosion of misinformation that could result from these new, unregulated tools.
Realistic fiction
There’s no cure-all, as we’ve seen since the advent of social networks and the tide of misinformation that followed. But it’s better than nothing because doing nothing is the surest path to informational chaos.
Informing the reader about the origin of content is a matter of minimum transparency. You can now generate photos that correspond to your precise wishes: I want a Ukrainian soldier carrying a wounded old man in his arms, with a burning building in the background. You’ll get this photo even if the scene didn’t exist as such.
All the elements of the photo will be real, taken from existing online sources, but the image itself is fictional. A bit like a Napoleonic battle painting. Labeling such a photo as “Artificial Intelligence” is a bare minimum.
AI has shaken the whole world, and the need to regulate it is obvious to all.
The objectivity of a photo, to stay with this example, is an old debate. It all depends on the framing, which can make the shot say different things. It also depends on the “decisive moment”, as Cartier Bresson used to say. With artificial intelligence, this debate is outdated; we’re in the realm of realistic fiction.
Most platforms are American
Will digital platforms accept the Commission’s request? That’s the big question. The Commission is addressing the dozens of digital companies that are part of the “EU Code of Best Practice Against Online Disinformation”. This is a voluntary initiative. But Twitter, one of the world’s leading networks, has just withdrawn from this European code. Its owner, Elon Musk, is reluctant to accept any constraints in the name of the First Amendment in the U.S. Constitution.
One of the difficulties, of course, is that most of the platforms are American, both legally and philosophically. Europe is certainly a huge market, but it only has the power to regulate, failing to produce the technology and companies that go with it. This is its great weakness.
The fact remains, however, that AI has shaken the whole world, and the need to regulate it is obvious to all. Yesterday, Hollywood’s actors’ union began negotiating with studios over the rights to use their AI-generated images in the future, when films can be produced without actors.
It’s just one of many examples of a wave that is shaking up, above all else, our relationship with truth and fiction. It’s staggering.