Terrell Russell: This Old Network

The fake is getting a bit too real.

We’re not ready.

This is going to get worse before it gets better.

This Generative Adversarial Network (GAN) creates, from scratch, a new face every time it is refreshed:

https://thispersondoesnotexist.com

Doing this requires some super-expensive and powerful hardware, but now there’s a website that bypasses that requirement: generating a fancy new face is as easy as refreshing your browser with the aptly-named thispersondoesnotexist.com. Every time you load the page, an algorithm generates a new human face from scratch.

The website uses an implementation of machine learning known as Generative Adversarial Networks, or GANs. These programs “learn” from a large number of training inputs—say, real human faces—in order to produce new examples. Thispersondoesnotexist.com uses code previously released by Nvidia researchers on GitHub.

And in the same week, we have seen surprisingly human text come from a different algorithm:

At its core, GPT2 is a text generator. The AI system is fed text, anything from a few words to a whole page, and asked to write the next few sentences based on its predictions of what should come next. The system is pushing the boundaries of what was thought possible, both in terms of the quality of the output, and the wide variety of potential uses.

When used to simply generate new text, GPT2 is capable of writing plausible passages that match what it is given in both style and subject. It rarely shows any of the quirks that mark out previous AI systems, such as forgetting what it is writing about midway through a paragraph, or mangling the syntax of long sentences.

Our old human brains are about to get lapped.

We need verified identity. We need provenance. We need education.

We need vigilance.

We need to greatly increase our collective ability to spot, label, and disregard unreliable information.

Tags: ai - machinelearning