If your Instagram is inundated in algorithmically created images of your friends, you aren’t alone. The photo editing software Lensa AI recently went viral after releasing a new avatar creation tool based on Stable Diffusion. Users shared their uncanny AI-crafted avatars (and the terrible misfires) in tales and posts.
What is the Lensa AI app?
The first way that many people have interacted with a generative AI tool is through Lensa’s entertaining, eminently shareable avatars. It’s also the first time they’ve paid for computer-generated art in Lensa’s case.
Many individuals are experimenting with Stable Diffusion for enjoyment or for research purposes because it is free to use. However, Lensa and similar services such as Avatar AI and Profile picture. Selling the processing cycles needed to process the prompts and generate the set of images is how companies like AI, to name a few, make money. That undoubtedly modifies the equation slightly.
Stable Diffusion’s free, open-source image generator serves as the foundation for Lensa, which serves as a middleman. Send 10–20 selfies to Lensa and pay $7.99 ($3.99 if you sign up for a free trial), and the software will work behind the scenes to create stylized photos in a variety of genres, including sci-fi, fantasy, and anime. Stable Diffusion can be installed on a machine, some models can be downloaded, and similar results can be obtained, but Lensa’s avatars are impressive and Instagram-ready enough that a large number of users are more than eager to pay for the convenience.
The development of AI image and text generators has been celebrated in the tech community this year, and artists have closely observed the developments. However, the average Instagram user is unlikely to have engaged in a philosophical discussion with ChatGPT or given absurdist instructions to DALL-E. The majority of people haven’t yet considered the moral ramifications of freely accessible AI technologies like Stable Diffusion, which have the potential to transform entire sectors if we let them.
There is no way for individual artists to opt out of being included in the training datasets. The vast collections of data are, in LAION’s words, “essentially indexes to the internet,” lists of URLs to images found throughout the internet along with the alt text that characterizes them. LAION is the nonprofit that originally developed the enormous datasets. You can file a removal request in accordance with the GDPR, Europe’s ground-breaking privacy law, if you’re an EU citizen and the database contains a photo of you with your name attached, but that’s about it. The horse is no longer inside the barn.
We’re still trying to figure out what this implies for artists, whether they’re large, copyright-conscious corporations or independent illustrators and photographers who get caught up in the AI modelling process. Some models that employ stable diffusion exacerbate the problem. Before a recent update, Stable Diffusion Version 2, anyone could create a named template intended to imitate a particular artist’s distinctive visual style and create new images at a rate that no human could match.
A interesting interview with Andy Baio, who co-founded a festival for independent artists, is posted on his blog and explores these issues. He talked to an illustrator who had found an AI model created expressly to imitate her work. My initial reaction was that having my name on this tool felt intrusive, she admitted. “… I wouldn’t have said yes if they had asked if they could do this.”
By September, Dungeons & Dragons artist Greg Rutkowski was concerned that his original work might be buried in the sea of algorithmic reproductions because his Stable Diffusion prompt was so widely utilized to create images in his meticulous fantasy style. “How about a year from now? Because of the abundance of AI art on the internet, I probably won’t be able to locate my work, Rutkowski told MIT’s Technology Review.
As more people come across these challenging challenges — and the existential threat they seem to offer — for the first time, those concerns, which have been expressed by many illustrators and other digital creatives, are resonating on social media.
“I am aware that many individuals have recently posted their Lensa or other AI portraits. In a well-liked tweet thread about Lensa, voice actress Jenny Yokobori penned, “I would like to encourage you not to do so or, better yet, not to utilise the service. Another expressed the concern that Riot Games artist Jon Lam had with AI-generated art. “When AI artists steal or appropriate our work, I don’t just see art; I see mentors and friends as well. I don’t anticipate you to comprehend.
Personally, I was sick over the weekend and confined to my house, where I squandered more time than normal mindlessly scrolling through social media. My Instagram stories were a flurry of cheap, flattering digital graphics. The appeal of polling your friends to determine which self-portraits are the most accurate representations of you (from my experience, most are) and which are hilarious mutations that could only be created by a computer performing its best human impersonation, shows that Lensa has undoubtedly discovered something special there.
Some friends, primarily artists and illustrators, resisted and insisted that everyone find their own artist to pay. It’s difficult to criticise the creative people in my circles who received compensation.
For better or worse, it’s truly impressive what the current generation of AI picture generators can accomplish, especially if you’ve only recently tuned in. We’ll all start paying attention soon. I downloaded Lensa and tried the app out for the purposes of tale research and vanity. In the past, in 2016, I had only ever paid once to have an artist create a single image for my profile. I now had a set of 50 epic avatars created from my most authentic photos for less than $10, but these were extra me. Me appearing out of a graphic novel in a variety of futuristic jumpsuits, me dressed in purple robes like an extraterrestrial saint, me, me, me.
I can see the allure. Several acquaintances mentioned how the images made them feel, implying the gender pleasure of being perceived as they saw themselves. It’s all really interesting and at least that complicated, so I wouldn’t hold it against anyone if they wanted to investigate it. Although I enjoy using them, there are times when I wish I didn’t. They won’t be put to use by me.
When I remember to update my online store, I sell largely nighttime photographs of the night sky and mountainous scenery. This got me to thinking about my own artwork. I recalled a few of the prints I’ve sold and the work it took to get the pictures.
A five-hour backpacking trip up to a remote fire lookout in Washington with special authorization from the National Park Service was required for one of my favorite photographs. Many involved long, lonely hours spent by myself in the bitter cold tending to my tripod and watching the Milky Way, which spun over a dark horizon like the hand of a clock.
The AI models already have sufficient training data to accurately recreate images of one nearby hidden mountain location that only local nightsky photographers appear to be aware of. When I took pictures there three years ago, I had to compete for a campground and travel miles up a rough forest service road just to spend hours waiting in the dark.
