Have you noticed how many of your friends have suddenly become fairy princesses or space travelers?
Is your Instagram feed flooded with Renaissance-style paintings of people born in the 1990s? If that’s the case, you have a right to know what’s going on (and it’s not time travel).
Users have flocked to Lensa AI in the last week, an app that uses your selfies and artificial intelligence to create portraits in a variety of styles.
The app, developed by Prisma Labs, is generating images—and controversy.
What exactly is Lensa AI?
Even if you’ve never heard of Lensa AI, you’ve probably seen some of its work this week. It was the most popular iPhone app in the United States in Apple’s app store as of Wednesday.
Lensa looks at your selfies and makes unique portraits of you or anyone else whose photos you send it.
Is it necessary for me to pay for it?
You do. During a one-week trial period, you can get 50 avatars—10 images in five styles—for $3.99. (You can purchase a one-year subscription to Lensa AI for $35.99, saving 51% on additional avatars.)
Lensa’s checkout page says, “Magic Avatars use a lot of computing power to make amazing avatars for you.” “It’s pricey, but we tried to keep it as low as possible.”
Fair warning: Prices have fluctuated as the app has grown in popularity and may have changed since this article was published.
Read: What Is ASMR?
How does it function?
You’ll upload a slew of selfies after downloading the app. (If you don’t want to pay money to get back a mess of images with phantom phalanges hanging from your mouth, don’t include any where your hands are touching your face.)
Choose a gender—male, female, or other—and walk away from your phone for around half an hour, and presto!
Your face, or something similar to it, has been stretched and squeezed across a suite of 50 to 200 A.I.-generated images with themes such as “cosmic,” “fairy princess,” and “anime,” depending on the package you purchase.
The Science Behind it?
Lensa employs stable diffusion, a technique in which all the horses in a barn spread out to give each other some space.
Stable Diffusion is a “really powerful” A.I.-based image generator, according to Subbarao Kambhampati, an Arizona State University professor in the School of Computing and Augmented Intelligence.
Stable Diffusion, like WALL-E 2 and Midjourney, uses image prompts (like your selfies) and text prompts (like “fantasy,” one of Lensa AI’s categories) to generate high-quality images that can get “trippy,” according to Dr. Kambhampati.
“It’s showing you pictures that nobody took; it’s just stitching them together from all the other images it’s seen.”
It’s fantastic! It’s amusing to imagine yourself as a painting, an anime character, or a little woodland elf with two very different-sized eyes and only one hand, even if you have two in real life. (Not every image will be perfect.)
Some Lensa users, especially transgender and gender nonconforming people, are finding that making things with Lensa makes them feel good about their gender.
I believe some people are upset about this. Should I be upset about this as well?
Some users are annoyed by the pictures Lensa AI makes for a number of different reasons.
The first is that many users claim the art is sexualizing them. When we tested the app, several of the images we got after uploading selfies and choosing “female” were of the whole body, even though users are told to only upload close-up selfies.
One image showed our avatar wearing a metallic bikini à la Princess Leia in “Return of the Jedi,” while another showed only half a face atop a scantily clad body.
According to Prisma Labs’ FAQ page, “occasional sexualization is observed across all gender categories.”
As a result, it’s relatively simple to use the app to generate lewd images of anyone you want.
This week, Tech Crunch was able to make topless avatars of famous people by putting pictures of an actor’s head on pictures of topless bodies.
Prisma Labs’ chief executive and co-founder, Andrey Usoltsev, told TechCrunch that using Lensa AI to engage in “harmful or harassing behavior” was a violation of the company’s terms of service.
What do users think of the app?
Yasemin Anders, 29, was inspired to use Lensa AI after seeing it on social media, and she was especially thrilled to see herself transformed into an ethereal fairy.
She was ultimately disappointed to see that Lensa had slimmed down her body and given her a slimmer face and neck. “If it’s smart enough to turn you into a fairy or a manga figure, you’d think there’d be some kind of smart enough software behind it to detect fat people as well,” Mx said.
Anders is a marketing professional who lives in Berlin.
“Even if I imagine myself as a fairy, like an idealized fantasy version of myself,” she continues, “I still want to look like myself.” “And I’ve always been overweight.” I was a chubby kid.
I was a fat teen, and now I’m a fat adult. So why would I want to imagine a perfect version of myself that doesn’t even resemble me? ”
Other users have reported that the app lightens or whitens their skin and significantly alters their facial features, which Dr. Kambhampati said was a common issue with generative A.I. tools, such as Snapchat’s face lenses
“If the majority of the images you fed the system were of white faces, it’s not surprising that when it tries to make a supposedly ‘better’ image, it just makes it whiter,” he explained.
Has Prisma Labs addressed these concerns?
In a Twitter thread, Prisma Labs stated that “A.I. “Lensa will not replace digital artists,” he said, rejecting the charge that it was stealing artists’ work.
“The AI learns to recognize the connections between the images and their descriptions, not the artworks,” according to the company.
“Just like movies didn’t kill the theater and accounting software didn’t kill the job, AI won’t replace artists, but it can be a great tool to help them.”
Is there any concern about privacy?
“I doubt that the entire business model is, ‘Give us $10 or $15 and we’ll send you an A.I. back.’ glam shot,'” said Jen King, a privacy and data policy fellow at Stanford University’s Institute for Human-Centered Artificial Intelligence.