This opinion was made by scientist Jan Goodfellow from Google Brain. He said that AI approaches the level when it can easily falsify any photos or videos. As soon as this happens, visual news content will be greatly depreciated-everything that used to serve as irrefutable evidence can cease to inspire confidence, writes Technology Review.

Goodfellow says that now he is engaged in similar projects in Google Brain. For the most part, these are generative and adversarial networks - a concept that he himself described in 2014. These are special machine learning algorithms that are built on the rivalry between two neural networks. One recreates various models, and the second verifies their authenticity. In the course of such rivalry, the models are getting more and more perfect. And already now such networks are capable of creating photographs, the authenticity of which a person is not able to establish.

Fakes were created before, but for a really high-quality counterfeit, huge resources were required. AI can drastically change the situation. Algorithms make the process of fraud available and much faster. At the same time they do not stand still and constantly study how to make the result of their work even more reliable. Because of this, Goodfellow believes that soon in the media and the news environment there will be a huge number of such high-quality fakes that will be fooled by a very large number of people.

It is difficult to fight against this, therefore, in his opinion, people will simply become more skeptical. Someone at all will cease to consume a photo and video content, because completely will lose to it trust. "Historically, we rely on video as evidence that the event really happened," he said during a speech. But with new technologies, you may have to change something.

According to Goodfellow, there were times in history when people coped without news with video and photo materials. Perhaps we will have to return to these times.

As soon as AI took up the editing of photographs, immediately there were fears that one day it would come to making fakes. While assumptions were being constructed, the algorithms were taught to forge not only images, but also the voices of specific people. So under threat not only visual news, but radio.