“Deepfakes” have startled the world lately. The imaging technique allows for paintings to appear as though they are talking. The latter is generated from a single image. For instance, the technique was tested on Mona Lisa, resulting in a “living portrait” wherein she appeared to laugh and then mouth words. The seemingly realistic videos and audio generated stems from the use of machine learning. The invention was developed by Samsung researchers operating out of Moscow. The credible videos are generated from a single image hence with just a few photos of the real faces, the results could look even more real. The creators described it as “photorealistic talking heads.” Though, it gets creepier.
The researchers defined the new technology as “puppeteering” which relates to a method where invisible strings manipulate a face. Indeed, the discovery could easily serve as the basis of a horror movie. Relatedly, concerns about the malicious use of the technological advancements have been raised. Folks are debating on whether “deepfakes” could be used to toy with politics and worldly affairs if one was cleverly crafted to mimic a trusted public figure. This would be very similar to a Black Mirror episode and may lead to greater damage than progress.
Moreover, researchers affirm this technology will soon make it hard to distinguish the real from the fake. At this point, the issue wil be to look at contextual cues in a robust way to help determine whether or not it is a fake. Thus, a more robust way of fact-checking your sources. Despite this, it remains really scary what the future has in store for such technology.