Fake news is a phenomenon harder to distinguish from real news not only for the content of the articles and their viral effects on social networks but also because its easier to create realistic and attractive fake videos and photos now. Further, fake news is a very damaging practice because they influence people’s behaviors, results of elections, and scams. Imagine what would happen if artificial intelligence was able to create fake but realistic video clips of people from a single photo of them. Samsung just developed a new AI which can do this.
Inside an artificial intelligence lab located in Russia, Samsung‘s AI was able to fabricate a fake video clip of people from just a single photo or even a painting. And it even looks quite realistic. One of the video clips published shows Leonardo Da Vinci’s Mona Lisa talking and moving, and the painting was the only image of the Mona Lisa used to make the video as it is the only one existing! Until now, the creation of deepfake videos needed a lot of images in order to provide a realistic result, but thanks to Samsung, just one photo is needed.
The new creation by Samsung is called “realistic neural talking heads” and the word neural is not just a stylistic choice. Neural networks are a type of machine learning that is able to mimic the human brain, and they share something in common with the new Samsung’s algorithm. In order to create the video clip, the AI watches a lot of videos to learn how human faces move, then it applies all it has learned into a single photo.
There are a lot of sectors which could benefit from such a technology, but we hope that it’s not put to bad use like creating fake viral videos.
UP NEXT: Xiaomi Mijia Bicycle Pump Review: The First Portable Pump
