Artificial Intelligence Helps In Creating Realistic Video Fakes - Alternative View

Artificial Intelligence Helps In Creating Realistic Video Fakes - Alternative View
Artificial Intelligence Helps In Creating Realistic Video Fakes - Alternative View

Video: Artificial Intelligence Helps In Creating Realistic Video Fakes - Alternative View

Video: Artificial Intelligence Helps In Creating Realistic Video Fakes - Alternative View
Video: AI Makes Near-Perfect DeepFakes in 40 Seconds! 👨 2024, April
Anonim

Edited videos appear on the Internet in which people say certain words (although they did not actually say them) or perform some actions (although they did not actually do them). Thanks to new developments based on artificial intelligence, the quality of such video fakes will increase significantly.

Today, technologies for creating fake videos are not limited to the movements of a person's lips and face in the original frames - an artificial intelligence-based system is able to prepare a realistic video, in which an edited digital image of a person can move his head in different directions, blink and show emotions. All actions and words of the actor who is filmed in the original video will be transferred to the fake video.

The development team compared the new algorithm to existing video and photo manipulation systems that Facebook and Google are involved with. The new algorithm surpassed existing techniques, and the participants in the experiment could hardly determine the authenticity of the videos.

The researchers were partly funded by Google. The creators of the system hope that the new technology will improve virtual reality. AI only needs to study a few minutes of the source material to start creating a modified video. Therefore, the researchers argue that high-quality video editing software will become more affordable.

The developers understand that their invention may raise some concerns in the community.

Researcher Justus Tees told The Register that he was worried about the ethical side of the issue. The results of the study were published to alert people to advances in manipulation techniques, Tees said.

Scientists once again "warned of the danger", but continued to improve the suspicious technology. In the text of the study, the ethical issue was mentioned only in passing: the developers believe that someone should improve the watermarking system or come up with another way to detect video forgeries.

The researchers themselves do not want to deal with this issue, as they are busy working on improving the tool for creating fake videos.

Promotional video:

Anton Komarov