Have you noticed that many of your friends are suddenly fairy princesses or space travelers? Is your Instagram feed overrun with Renaissance-style paintings of people who were definitely born in the 90s? If so, you’re entitled to an explanation of exactly what’s going on here (and it’s not time travel).
Over the past week, users have flocked to Lensa AI, an app that uses your selfies and artificial intelligence to create portraits in a variety of styles. Created by the company Prisma Labs, the application generates images – and controversies.
What is AI Lensa?
Even if you haven’t heard of Lensa AI, you may have seen its work this week. It was the most popular US iPhone app in the Apple App Store on Wednesday. Lensa takes your selfies, studies them, and produces original computer-generated portraits of you – or anyone whose photos you feed it.
Do I have to pay for this?
You do. Right now, you can get 50 avatars – 10 images in five styles – for $3.99 for a one-week trial period. (For $35.99, you can subscribe to Lensa AI for the year, which gets you 51% off future avatars.) “Magical avatars consume enormous computing power to create amazing avatars for you,” according to Lensa’s checkout page. “It’s expensive, but we’ve made it as affordable as possible.” Fair warning: prices have fluctuated as the app has become more popular and may have changed since this article was published.
How it works?
After downloading the app, you will download a bunch of selfies. (Do yourself a favor and don’t include any places where your hands touch your face unless you want to pay to pick up a mess of pictures with phantom knuckles hanging from your mouth.) Select a gender – male, female, or another – and walk away from your phone for about half an hour, and when you come back, presto. Your face, or something like that, has been stretched and pressed over a suite of 50 to 200 – depending on the package you purchase – AI-generated images with themes such as “cosmic”, “fairy princess” and “animate”.
Ok, but how is it going really work?
Lensa uses Stable Diffusion, which is when all the horses in a stable spread out to give themselves some space. Just kidding: Stable Diffusion is a “really powerful” AI-based image generator, said Subbarao Kambhampati, a professor at Arizona State University’s School of Computing and Augmented Intelligence. Similar to DALL-E 2 and Midjourney, Stable Diffusion uses image prompts (like your selfies) and text prompts (like “fantasy”, one of Lensa AI’s categories) to generate high-quality images that sometimes become “trippy”, Dr Kambhampati said. “He shows you pictures that no one has taken; he’s just able to put them together from all the other images he’s seen.
Cool.
It’s cool! It’s fun to see yourself rendered as a painting or an anime character or a little wood elf with two very different sized eyes and one hand, even if you have two IRLs. (Not all images will be perfect.) Some users, especially transgender and gender-nonconforming people, even find that using Lensa to create can deliver a sense of gender euphoria.
I think some people are crazy about this? Should I be mad about this, too?
There are several reasons why some users bristle at the images that Lensa AI spits out. The first is that many users report that the art sexualizes them. When we tested the app, several of the images we received after uploading selfies and selecting “female” included full-body renders, despite users being specifically instructed to upload close-up selfies only. An image showed our avatar in a metallic Princess Leia bikini in “Return of the Jedi.” Another included only half a face on a scantily clad body.
Prisma Labs states on its FAQ page that “casual sexualization is seen across all gender categories”.
Thus, it is quite simple to use the application to create obscene images of whoever you want. This week, Tech Crunch was able to create topless avatars of celebrities using images of an actor’s head mounted on topless bodies. “Turns out the AI is taking these Photoshopped images as permission to go wild, and it appears to be disabling an NSFW filter,” Tech Crunch reported. Now imagine that same experience, but through the lens of someone looking to do revenge porn. It gets murky in a hurry.
Andrey Usoltsev, chief executive and co-founder of Prisma Labs, told TechCrunch that using Lensa AI to engage in “harmful or harassing behavior” is a violation of its terms of service.
What are app users saying?
Yasemin Anders, 29, decided to use Lensa AI after seeing it on social media and was especially thrilled to see herself recreated as an ethereal fairy.
She was, however, ultimately disappointed to see that Lensa had given her a slimmer body and slimmer face and neck. “If it’s smart enough to turn you into a fairy or a manga character, you’d think there’s some kind of software smart enough to detect fat people as well,” Mx said. Anders, who lives in Berlin and works in marketing.
“Even though I imagine myself as a fairy, as an idealized fantasy version of myself, I would still like to look like myself,” she added. “And I’ve always been fat. I’ve been a fat boy. I’ve been a fat teenager, and I’m a fat adult now. So why would I want to imagine an ideal version of me that doesn’t even look like me? ?”
Other users report that the app makes their skin lighter or whiter and drastically changes their facial features, which Dr. Kambhampati said was an ongoing issue with generative AI tools, including lenses. Snapchat facials. “If most of the images you fed into the system were white faces, it’s no surprise that when it tries to create an image that looks ‘better’ it just makes it look whiter,” he said. -he declares.
What do the artists think of Lensa?
Stable Diffusion was formed on the creations of many artists who did not explicitly consent to the use of their work for the benefit of Prisma Labs, said Dr. Kambhampati, adding: “If van Gogh were alive today, you would probably need to pay van Gogh. license fees to have your photos in the style of van Gogh. That’s not what happened here.
“For a lot of people, having our art stolen, they don’t see it as something personal – like, ‘Oh, well, you know, it’s just a style; you can’t copyright a style,” said Jonathan Lam, a screenwriter who works in video games and animation. “But I would say that for us, our style is actually our identity. This is what distinguishes us from each other. This is what makes us marketable to customers. Mr. Lam, who lives in Vancouver, British Columbia, is one of many artists who have since spoken on social media about the app.
He also noted that some of the works generated by Lensa include what appear to be renderings of artists’ signatures. (Several of our own Lensa-generated images had squiggles where a signature might go over an IRL painting.) “All these tech enthusiasts say these generators are creating something new, but if the artist’s signature is still there , it’s not something new,” Lam said. “It just generates something based on the data that fed it.”
“I think the general public is under the assumption that it’s a program that just learned to draw on its own in a vacuum, and they don’t realize the bigger implication of data mining,” he added. “When you start to see these things get monetized and real people being exploited and abused, it’s pretty scary. And it’s even scarier when it’s disguised as a pretty app.
Has Prisma Labs addressed these concerns?
Prisma Labs wrote in a Twitter thread that AI “will not replace digital artists” and pushed back on the characterization that Lensa was stealing artists’ work. “AI learns to recognize links between images and their descriptions, not works of art,” the company wrote. “As cinema did not kill theater and accounting software did not eradicate the profession, AI will not replace artists but can become a great support tool.”
Are there any privacy issues?
“I doubt the whole business model is, ‘Give us $10 or $15 and we’ll send you back a glamorous picture of AI,'” said Jen King, privacy and data policy officer at the Institute for Human-Centered Artificial Intelligence at Stanford University. Lensa AI’s privacy policy states that “facial data” is deleted within 24 hours of processing and is not used to identify an individual user – but it also states that your photos and videos may be used to further train users. Lensa algorithms. Would Dr King use Lensa AI? “Nope.”
It Happened Online is a column in which we explain very particular information made possible and amplified by social networks.
#selfies