AI-Generated Feet Pics Are Here—But Not (Really) for the Reason You Think

"Ha! If there are people out there that can get off to these, then I salute them."

This Foot Does Not Exist
These feet ~literally~ do not exist in real life. Courtesy of MSCHF/This Foot Does Not Exist

Out of all things that artificial intelligence (AI) can do to benefit the good of the humanity, generating fake feet photos has got to be… on the very top of the list?

Sign Up For Our Daily Newsletter

By clicking submit, you agree to our <a href="http://observermedia.com/terms">terms of service</a> and acknowledge we may use your information to send you emails, product samples, and promotions on this website and other properties. You can opt out anytime.

See all of our newsletters

Rest assured humanity, your AI-produced fake feet photos are here.

SEE ALSO: Sex Workers Are at the Forefront of the Fight Against Mass Surveillance

MSCHF is a Brooklyn-based creative collective that spews out a constant stream of new, interesting digital/real-world projects. They’re latest creation is This Foot Does Not Exist (TFDNE), a bot that you text to receive GAN (Generative Adversarial Network) generated foot photos.

Here’s the rundown: You text the phone number listed on the TFDNE website (646-760-8955). And within a couple of minutes, you get a photo texted to you that looks something like this.

AI-generated foot photo

Or this:

AI-generated foot photo

Or even this:

AI-generated foot photo

This Foot Does Not Exist text messages.

“Do you like it?” questions the bot after sending the AI-generated foot photo. 

To refresh, GAN is class of machine learning, in which two virtual bots contest with each other to better results; in this case, feet photos. Given a training set of images, the bots learn to generate new data with the same set of statistics as the training set. The AI world has Ian Longfellow to thank for inventing GAN. 

In the case of TFDNE: “It’s a GAN that’s been trained on 10,000 images of feet,” explained MSCHF’s Samantha Thompson. “A GAN uses two neural nets; in this case, one tries to create images of feet, and the other judges whether the outputs are real or not. So, the process essentially competes with itself and, in this manner, improves over time.”

So, why is there a demand for this? Is it for the foot fetish community, who is often after new and different looking feet… that the real world just can’t conjure? 

“Ha! If there are people out there that can get off to these, then I salute them,” said Thompson. “TFDNE has a few different appeals, depending on the angle. There’s a definite tech audience for people who like to follow unusual uses of GANs and image-creation. One of the things TFDNE also uniquely offers is the experience of asking for and receiving foot pictures. It’s a gratifying thing to ask for and receive… Our bot simulates this—albeit in a humorous manner.”

And then you get this:

AI-generated foot photo

And this:

The general reaction to TFDNE from live humans?

“Some people love them, some people are horrified by them,” said Thompson. “A handful of people have expressed concern that we might eventually drop the bottom out of the foot pic market and reduce prices to $0, but for the most part they’re just endearingly grotesque.”

Thompson also expressed her own feelings about the AI-generated feet pics: “They’re hot, hot, hot!”

Some of MSCHF’s previous creatively inventive projects include: Jesus Shoes (a customized Nike Air Max 97 filled with Holy Water), Netflix Hangouts (a Chrome extension that allows you to watch Netflix at work by simulating a Google hangout), Times Newer Roman (a wider version of the popular font that pads out text in essays), The Persistence of Chaos (a laptop infected with six of the world’s most dangerous computer viruses) and Man Eating Food (a YouTube channel featuring a man eating numerous types of different exotic celebrity food).

But in terms of foot photos generated by machine learning, “AI is going to continue to interface with human image production, and it’s only going to get better. There have been scores of distasteful implementations already, even at the limited level of finesse available so far,” said Thompson.

MSCHF’s overall takeaway with their TFDNE project is both a statement on the tech and our culture. Their aim is to represent a a smarter and more thoughtful intersection of GANs and (porn-adjacent) images.

“As deepfakes get better, as GAN-made faces, bodies, people become more and more convincing, we’re all going to need to get smarter about this kind of media as a culture,” concluded Thompson.

“We hope that the research into the cultural function and production of foot pics that we’ve conducted with TFDNE sets a precedent for an incrementally greater degree of thoughtfulness, despite the absurdity of its subject.”

MSCHF releases new items every 2nd and 4th Tuesday of the month.

AI-Generated Feet Pics Are Here—But Not (Really) for the Reason You Think