On May 2019, researchers of the Samsung AI Center and the Skolkovo Institute of Science and Technology, based in Moscow, published a paper presenting a brand-new Samsung app. The futuristic algorithm can create short videos using photos of a person. Although several shots from different angles help produce a more realistic animation, the app is able to generate a movement using only one picture, or even a painted portrait! A mind-blowing achievement, evoking the technologic excesses of the TV series Black Mirror.
Deep fakes and artificial intelligence
The system works with a process of face recognition, which can then be manipulated. This algorithm is based on “landmark features”. It recognises the models’ common characteristics and is easily able to associate them with another face. The first results, published in a video, were produced thanks to an online library called VoxCeleb, sheltering more than 7000 images of famous people.
According to researchers, this new technology might help developing video games and special effects. However, resembling deep fakes (a technique based on artificial intelligence, used to combine and superimpose existing images onto source images – a person’s face, for exemple, ed.), this new algorithm raises questions. How do we combine artificial intelligence and ethics? Will the evolution of technology encourage the development of fake news? Could one get their identity stolen?
Source: Cornell University