Menu

Search

  |   Technology

Menu

  |   Technology

Search

AI Creates Realistic Fake Photos Of People Who Don’t Exist

Artificial intelligence has now advanced to a point where it can even create photographs of people who don’t exist. These images are apparently so realistic that they can’t be identified as fake without some sophisticated machinery. The development does come with unique possibilities, especially with regards to the entertainment industry, but it’s undeniably eerie as well.

The AI was able to generate such images due to an algorithm called the generative adversarial network (GAN), which was used by GPU giant Nvidia, Futurism reports. This particular GAN is leagues better than what other companies have managed to create in the past, which were fairly powerful on their own. Earlier examples could turn doodles into realistic representations of faces.

As to how GAN works, it basically pits two types of neurons against each other. One does the creation while the other challenges what was created. In this way, the results improve. Using the database at CelebA-HQ, which stores pictures of celebrities and other figures, Nvidia fed its GAN with images to create photos of people who do not exist.

In the paper that the team behind the project published, it was noted how the adversarial system quickly created images that were practically indistinguishable from real people. While Nvidia is nowhere near the first company to develop a machine capable of producing realistic photos of fake people, its project is certainly one of the most efficient.

When combined with the work of other companies that replicate human speech, human skin, and even human expressions and emotions, this particular development creates a particularly disturbing scenario. Machines that can walk, talk, express, and look like everyday people would no longer be so farfetched.

On the other hand, this fake image generation technology could have huge benefits in the realm of video games. With advanced graphics being one of the main draws to more powerful systems, Nvidia’s own technology could pave the way for video games that look more realistic than ever.

  • Market Data
Close

Welcome to EconoTimes

Sign up for daily updates for the most important
stories unfolding in the global economy.