Should we fear these false images created by artificial intelligence? The Franceinfo Talk debate

The pope in a down jacket or at an LGBT+ demonstration, Donald Trump arrested or Emmanuel Macron picking up the trash: more and more very realistic but yet false images are circulating on social networks. They were created on demand by artificial intelligences such as, for example, Midjourney or Dall-E. So, how to detect a false image and above all what questions does it raise for information professionals? How to avoid deviations?

Manon Mella receives Gerald Holubowiczjournalist and media consultant, synthetic media specialist as well as Thomas Pontillonjournalist at franceinfo.

How to spot fake photos?

Artificial intelligence is becoming more and more efficient. However, there are still a few small details that escape the technology. The AI ​​can sometimes err in shades of color and contrast or generate asymmetries and bodies in odd positions. However, the platforms that allow these images to be created are improving rapidly.

Thus, the question of false images generated by artificial intelligence becomes ethical. How to supervise or even limit the dissemination of these “fakes”? If the end of free trials reduces the number of users, it is not enough to prevent the drifts that AI can cause. So, should we restrict the development of algorithms or prohibit the dissemination of these images?

Join us from 6 p.m. Monday to Friday on the franceinfo Twitch channel to participate in Manon Mella’s Talk. An hour of exchange, clarification, and debate around topical and social issues.