US technology company, NVIDIA, has published a video showcasing images of human faces, created entirely by artificial intelligence (AI). These lifelike images, which are referred to as “deep fakes” are almost impossible to tell apart from real photographs.
Created using an algorithm and 70 000 photos of people of all ages and ethnicity from the image hosting site, Flickr. The computer studied different markers on the images, like face shape, eye color, hair color, and many other aspects to generate each new image.
NVIDIA has only been operational for four years but in that timeframe, considerable progress has been made. The technology called Generative Adversarial Networks (GANs) is so accurate that most people have trouble distinguishing between the AI created
NVIDIA has applied the StyleGAN to pictures of cars, bedrooms, and cats as well, although the accuracy of the images created is not of the same standard as that of the faces.
The possibilities the StyleGAN has created has also sparked debate amongst the AI community, as well as, on social media as to whether the technology could be utilized for malicious purposes for fake news and propaganda.