AI-Generated Identity Can Now Make You a Beautiful Young Girl

Beware! The pretty girl you follow on Insta might be a fifty-year-old biker using AI!

It's about time we stop believing what we see on the internet! With the auto-generation power of AI and machine learning, internet users are posting images of themselves that are 180 degrees opposite of how they actually look. Now, some people are calling this major act of Catphishing to be art!

Let Me Tell You a Story

@azusagakuyuki is a young Japanese motorcyclist with long hair, a delicate chin, and 33,000 Twitter followers. There, she posts pictures of herself in a biker shirt, posing in front of her gleaming red-and-blue Yamaha Telkor on dirt roads and hilltops and misty beaches. She's beautiful, adventurous, and envy-inducing.  But one day, she accidentally posted a picture of her bike on Twitter that captured her reflection in the rear-view mirror. The reflection was of a middle-aged man–because the woman in the photo was actually a 50-year-old man named Soya who transformed his face using a machine-learning-powered face tune app. "No one will read what a normal middle-aged man, taking care of his motorcycle and taking pictures outside, posts on his account," Soya told the Japanese TV program Getsuyou Kara Yofukashi. That said, happily, his fans responded mostly positively to his late-in-life, accidental gender reveals.

Now Comes the Question of Ethics

How much worse is not too bad? This is a question that is being asked by the netizens in light of such an incident. According to them, if your end goal isn't money or sex, but instead, merely, to be a digital content creator, there is no harm in presenting yourself in some ways that would bring in more followers. However, putting a filter on an image of yourself is not a new thing. But turning a 50-year-old man into a 20-year-old young woman will take some pretty heavy ML algorithms to work. The recent advances in image-generating machine learning will blow your mind, and perhaps pull you deep into a vanity tailspin. You can make yourself more "stunning," give yourself a nose job, change your hairstyle, clear your acne and wrinkles, and straighten your teeth. You can do this all in a way that looks shockingly "natural," as though all you did was snap a photo in perfect lighting conditions.

Generating Fake People with Machine Learning

Amidst all this, the line between simply elevating your facial features to completely generating a fake face is blurring pretty fast. Machine learning for image generation started gaining traction back in 2014 when researchers from the University of Montreal showed they could use what's called a "generative adversarial network" (GAN for short) to generate blurry, black-and-white human faces. In 2018, Nvidia showed they could generate infinite faces using an improved-upon technique and the images were in color. None of these people are real, but it's almost impossible to tell that unless you know what to look for. You can click through an endless number of these generated faces at thispersondoesnotexist.com. What's more, Nvidia's method allowed researchers not only to generate faces, but also to tweak those faces along various visual axes, like age, skin color, gender, and density of facial hair.

Can It Be Put to Good Use?

Creating a fake identity with ML advancements gives internet users a chance to anonymize their identity. Besides the catfishing dangers, there are also some examples where these algorithms can be put to good use. Last year, the filmmaker David France used an AI-powered tool like Anonymizer in his documentary "Welcome to Chechnya." The film chronicles violet anti-gay and lesbian purges in Chechnya, and in it, France wanted to include real interviews with gay and lesbian Chechens who were fleeing the region. For this, he needed a way to shield their identities. So, he turned to machine learning. The interviewees could be seen on screen without ever really being seen on screen.

The Advancements are Endless

Today, ML can be used not only to change someone's appearance but also their voice. A recent documentary called "Roadrunner: A Film About Anthony Bourdain" used machine learning to create a voiceover for the film in the chef's own AI-simulated voice, posthumously. All these advances lead us to some pretty deep psychological behavior of people to be inclined towards creating fake identities and the new age of AI and ML is supporting it vehemently.

More Trending Stories