Sunday, May 5, 2019

Don't believe your eyes - this person does not exist

In a world full of fake news it's time to embrace the concept of fake people. An artificial intelligence system called StyleGAN developed by Nvidia is used on a site called to generate an infinite number of portrait photos of people who do not exist (watch the video below to see how it works). Just go to the site and simply reload the page to generate a new portrait. Some are extremely convincing though sometimes there are flaws that make you suspicious, such as strange skin features or ears that don't quite fit. However, I don't think we are meant to study the faces for any length of time, we will probably just see them as a gallery of "satisfied customers" in an advert or concerned citizens in political propaganda.

The phenomenon is described in an article from CNN, These people do not exist. Why websites are churning out fake images of people (and cats). Of course if you can generate fake people you can also generate fake cats, fake home interiors, fake cars and much more. There are many valid uses for a tool like this.

GANs-produced fakery can be fun — if you know what you're looking at — and potentially big business. A startup called Tangent, for example, says it is using GANs to modify faces of real-life models so online retailers can quickly (and realistically) tailor catalog images to shoppers in different countries rather than using different models or Photoshop. A video game company could use GANs to help come up with new characters, or iterate on existing ones.

There are also, of course, lots of less honest applications of this technology and that's what makes it scary. At least by letting us see these applications they make us aware that this technology exists and can be more careful in the future but once it is perfected how will we ever be able to tell the difference between reality and fiction? I also wonder how you can decide the copyright for these images since they are mashups of a vast number of photos of real people. Maybe it's time to stop saying things like "I'll believe it when I see it."

No comments:

Post a Comment