There’s An AI-Powered Website That Creates Non-Existent Human Faces

By  |  February 17, 2019

If you thought human beings were the only ones with the ability to make other human beings, well you are wrong. Thanks to special efforts by Nvidia, who took advantage of an algorithm created by Googler Ian Goodfellow, called generative adversarial network (GAN).

GAN came to light in 2014, and it is an algorithm that consists of two neural networks: a generator and a discriminator. The generator creates an image of an object, and the discriminator is supposed to determine whether the image is fake or not. If the generator is able to fool the discriminator, then the image is passed as a ‘genuine sample’.

Nvidia’s team has been working on this technology and released their own version of GAN, StyleGAN where they trained the AI, using real human pictures collected from Flickr, to generate non-existent human faces all on its own.

This AI is accessible through and as per its developers, it is able to build entirely new human faces from scratch by toggling different styles, colours and genders. Here are a bunch of faces created by the AI:


The team chose to target human faces because apparently, they are the hardest to achieve, as the human brain is quite capable of recognising when a human face is not so “human”. As good, and creepy, as the AI is, it doesn’t always work out and there are instances where the face generated is just scary:

StyleGAN Fail

There are various potential uses for the algorithm including generation of human personas in games and Virtua Reality worlds and Phillip Wang, the website’s developer, says that he hopes his website will bring attention to the technology.

Feature image courtesy:

Gadgets Africa is a part of Wee Media Group. This article was originally written by Saruni Maina.

Most Read

The FTX Disaster Undermines The Hopes Of Africa’s Peculiar Crypto Scene

The rise of the cryptocurrency industry in Africa, however a fringe endeavour, is

“Banking The Unbankable” Kindles Banking-as-a-Service In African Fintech

Despite attention to the challenges surrounding financial inclusion, an estimated 57 percent of

Kenya’s Mobile Money Agents Aided Financial Inclusion—Now They Face Exclusion

In Ruiru, a town in Kiambu County which sits within the greater Nairobi