Recently, a picture got viral on the internet with a story of a 22-year-old girl, and how brutally she was murdered. Rest in peace!
Cracking the suspense, the story is fake. Surprising you more, this girl doesn’t even exist. Photoshop? Oh no, this photograph has been generated by Artificial Intelligence. Hold your breath, there’s a lot more to come yet. “This Person Doesn’t Exist” is a website that provides you with chains of artificially generated faces that starts off from a few source images. If you’re doubting the case, you can visit the website yourself and with each click, you’ll witness a new stranger face right in front of you.
Initiated by a software engineer from Uber, Philip Wang, along with the research work by a chip designer, Nvidia, this site now generates tons of fake faces. Getting an insight into the programming behind, the algorithm has been made and tested on many real images multiple times and trained on large data sets. The final step is the implementation of ‘Generative Adversarial Network’, a type of neural network algorithm.
Mentioning in a Facebook post, Wang wrote, “Each time you refresh the site, the network will generate a new facial image from scratch”. Furthermore, while giving a statement to Motherboard, “Most people do not understand how good AIs will be at synthesizing images in the future.”
Another acknowledgment for the AI framework for the site goes to a researcher named Ian Goodfellow. The algorithm has been made open source to be used across the world and can be customized as per the application it is intended for. Initially, the project aimed at generating human faces but following the same trend, it can be applied for anime characters, fonts, graffiti or mimic any source.
The algorithm is referred to as StyleGAN and it has led to many ideas assisting illustrators, designers and even in advertisements. This could also cut off costs involved in hiring models for publicity of products. Creative minds can lead to entire virtual worlds or could work on future models of gadgets based on present specifications. The whole sole idea revolves around merging the characteristics making sets of all possible combinations from the real sources available. Combining features give birth to faces that can be from different ethnicities, creeds, and geographical backgrounds.
On the other side, the negative aspects include copyright issues or the wrong usage of pictures leading to trust issues. Furthermore, it could be used for triggering political issues or targeting sensitive matters. In a nutshell, it’s a must to associate rules and carry strict monitoring of such realistic imageries, so that it adds to the benefits rather than gives a tough time.