A new website, ThisPersonDoesNotExist.com, demonstrates the technology’s power by using research to create an endless stream of fake portraits when you refresh the page.
According to The Verge, the algorithm behind the site is trained on a huge dataset of real images, but then uses a type of neural network known as a generative adversarial network (GAN) to produce the fakes.
“Each time you refresh the site, the network will generate a new facial image from scratch,” Philip Wang, a software engineer from Uber who built the site, wrote on Facebook.
The technology underlying the site was reportedly first invented by Ian Goodfellow. This version of the model was trained on human faces, but researchers are also experimenting with anime characters, fonts and graffiti.
Artificial intelligence-generated imagery — which does have positive uses for designers — can also be harnessed for deepfakes, in which peoples’ faces are pasted onto other photographs (sometimes pornography) without their consent.
In addition, the technology can be used to create political propaganda or influence campaigns, according to experts.
(Patrick T. Fallon/Bloomberg via Getty Images) A group of scientists backed by Elon Musk have designed a predictive text machine that is so eerily good its creators are worried about releasing it to the world. Designed by OpenAI, a non-profit artificial intelligence research organization backed by the eccentric billionaire, the machine can take a piece...
File photo – A 3D printed Android logo is seen in front of a displayed cyber code in this illustration taken March 22, 2016. (REUTERS/Dado Ruvic/Illustration) Getting online advertisers to stop tracking you over Android may not be as easy it seems. New research has found that over 18,000 Android apps can create permanent records...