AI art's hidden echo chamber is about to implode

Artificial intelligence creates millions of images a day, flooding the internet. But what happens when it starts to train on its own data?

- 

In the past year, art created by artificial intelligence (AI) has gone from being the subject of research papers to emerging as a niche fad, all the way through to becoming an internet-dominating tool producing millions of images a day.

To get to this point of ubiquity, however, all the Al models had to be trained. And their training involves a hugely comprehensive deep-dive of the internet, in which they scan billions of images along with the images' corresponding descriptive texts.

Not only does that raise some major ethical questions around copyright, it also begs one question for the future: what happens when the internet becomes flooded with images made by artificial intelligence?

As these models continue to train by scouring the internet, they will undoubtedly be trained on images they first created. Does that cause some sort of self-perpetuating loop of weirder and weirder images, or will nothing actually change?

THE LOSS OF CREATIVITY

"AI will eventually start training on its own work it's expected to happen. That will essentially lead to stagnation in creativity. They train on what is already on the internet, so they will copy what is popular out there," says Ahmed Elgammal, a professor of computer science at Rutgers University, in New Jersey.

"If you get into the cycle of feeding AI what's on the internet, which right now is mostly AI, that'll lead to a stagnation where it's looking at the sam thing, the same art style, over and over again."