Lensa is the app behind your friends’ new profile pics. Here’s what artists have to say about it.

It’s the tool transforming your friends into superheroes, astronauts, fairy princesses, and anime leads.

Lensa, an app that uses AI technology to morph photos into stylized images, has surged in popularity over the last few weeks, thanks to its new viral “magic avatars” feature. Users are posting their elevated selfies on TikTok, Instagram, and Twitter. The hashtag #Lensa has more than half a million posts on Instagram. The app was downloaded 1.6 million times in November, up 631% from October.

The spike has prompted the latest discussion surrounding art that’s created by artificial intelligence — which comes with its own pros and cons, artists say.

Lensa AI is one of several artificial intelligence image generators that lets users create and post artistic renderings of their own photos. The app has been around since 2018, but the “magic avatars” feature drew new interest.

The app can transform users’ photos into as many as 200 avatars. Lensa uses Stable Diffusion, open-source artificial intelligence that can generate images from text prompts, to make its renderings look like they were created by an artist — not a computer.

Here are the basics:

Ryan Evans, a New Jersey-based artist who works in Philadelphia, downloaded Lensa to try for himself.

“My narcissistic ass couldn’t stay away,” he said. “I had to spend the $3.99 to get 50 very bad illustrations of myself.”

Evans recognizes the allure tools like Lensa has.

“I think it’s just another fun tool, honestly,” he said. “It helps me generate things I would normally never be able to on my own. Things like Muppet creations, distorted 3D versions of my favorite characters, etc.”

Evans said he doesn’t use AI apps much these days, but called it “addictive” at first.

As noted by Refinery29, the majority of widely shared Lensa avatars have been posted by white influencers.

But the app has also become a source of affirmation for people — especially members of the LGBTQ community — who experience body dysmorphia or want to see a version of themselves that feels more authentic. It has also become a way for people to experiment with their image — like with different hairstyles — before committing to a real-life change.

Before Lensa’s recent jump in popularity, other artificial intelligence platforms, like DALL-E, had made their way into the mainstream.

DALL-E — which experienced its own trending period on social media earlier this year — lets users describe what kind of art they’re looking for in 400 characters or less, and the generator spits out its interpretation. On Twitter, users have posted some of their prompts and the weird and wonderful results.

From “a wise cat meditating in the Himalayas searching for enlightenment” to “A rabbit detective sitting on a park bench and reading a newspaper in a Victorian setting,” the possibilities seem endless.

Kingsley Spencer, a Jacksonville, Fla.-based creative director and designer, says that AI generators get smarter with use.

“Over the next 12 to 18 months, we’ll spend our time finding out the best ways to use these tools to our advantage while feeding their data sets the most honed, strategic inputs possible,” he said.

As the generators become stronger, Spencer said, they could potentially devalue their time and labor of creatives.

“Once their models are trained up enough to think like us — as in, the people usually hired to do creative thinking — they can sell products and services that more or less replace us,” he said.

Critics of Lensa and other apps that feed from open-source AI say the platforms used nonprofit loopholes to unethically access thousands of real images and photographs while profiting from the results. For example, DALL-E reportedly used stock art service Shutterstock to train its AI.

“A lot of people are unknowingly supporting that process of theft,” said Philly-based graphic designer and illustrator Nicole Saltzer. “It’s a hard battle because it seems fun and harmless, but I think this could be, and in a lot of ways already is, the beginning of some serious problems in all creative fields.”

To illustrate how quickly artificial intelligence services can crank out art, Evans used DALL-E to create an illustration for this article.

“I used a combination of detailed phrases, like ‘3D sculpture, digital menacing face, Frankenstein, neon green glowing, computer rendering photo-realistic on a white background,’ ” he said. “Programs like DALL-E tend to function better with longer, more convoluted phrases.”

After about 15 attempts stringing the terms together to get a style that matched his vision, Evans received a 3D rendering of an eerie green face with sunken eyes. He inputted that into Photoshop where he added some signature touches, including a binary Matrix-like code and colorful smiley face stickers, “to bring back the handmade touch only we can accomplish.”

The whole process took Evans a little over an hour. Without AI, he said, that time would have easily tripled.

Evans says it’s important for nonartists, especially companies looking to commission art, to consider the source of the product.

Saltzer thinks that as AI generators become more accurate, they pose more threats to creatives.

“A lot of artists were not so afraid of the impending AI takeover of visual art because [the results are] not very accurate,” she said. “They’re convincing but always just a little off. I think we’re seeing how accurate it is becoming and now we’re faced with some serious questions. It’s hard to argue with instant gratification at a low cost, but that won’t make it good and it certainly will not ensure it was acquired ethically.”

AI interest has also prompted privacy concerns.

Some worry about how their images will be used, mirroring the concerns raised in 2020 over FaceApp — a Russian app that used AI technology to create the then-trending photos of people aging.

According to Lensa, the photos are “immediately” deleted from its servers after the avatars are finished. The app’s privacy policy says it does not claim ownership but seeks permission to use user content.

At the end of the day, Evans says, “there will be companies and artists who use AI poorly. There will also be people who see it in a fun and interesting way. I just hope we’re all able to remember the difference.”