AI Faces: More Believable than Real People?

In this video, we’ll look at the results of a recent study where participants classified AI-generated faces as genuine two-thirds of the time, which was significantly more often than they did for photos of real people.

We might understand why people can be fooled by a good fake – but how can it be that AI-generated images actually score even higher on looking genuine than photos of real people?

How can that happen?

I’ll share evidence to show good reasons why.

Along the way, we’ll briefly review how generative adversarial networks like StyleGAN2 work, and we’ll also create some AI-generated faces for ourselves!

The video ends with a brief review of some legitimate uses for creating AI-generated “faces of people that don’t exist.”