Since Vincent van Gogh painted the swirling scene in 1889, countless artists have drawn inspiration from The Starry Night.
Today, artificial intelligence systems are doing the same, training themselves on vast collections of digitized artwork to generate new images that can be conjured up in seconds from a smartphone app.
The images produced by tools like DALL-E, Midjourney, and Stable Diffusion can be strange and otherworldly, but they are increasingly realistic and customizable. to what you imagined.
But while Van Gogh and other great painters long dead aren’t complaining, some living artists and photographers have an AI that creates images derived from their work. Some are starting to fight back against software companies.
Two new lawsuits have emerged, one of which was brought by Seattle-based photo giant Getty Images, alleging it copied and processed millions of copyrighted images without a license and takes aim at a popular image generation service.
Getty says it has initiated legal proceedings in London’s High Court against Stable Diffusion maker Stability AI for infringing intellectual property rights for the commercial interests of the London-based startup. rice field.
Another lawsuit filed in US federal court in San Francisco describes AI image generators as “21st century collage tools that violate the rights of millions of artists.” The lawsuit, filed by three practicing artists on behalf of other like-minded artists, also names Stability AI, San Francisco-based image generation startup Midjourney, and online gallery DeviantArt as defendants.
According to the complaint, AI-generated images “compete with original images in the marketplace. Historically, if a buyer wanted a new image ‘in style’ from a particular artist, the buyer would You must pay a fee to commission or license the image.”
Companies that provide image generation services typically charge users a fee. For example, after a free trial of Midjourney on his Discord chat app, users will have to purchase subscriptions starting at $10/month and up to $600/year for corporate memberships. His startup, OpenAI, also charges for use of the DALL-E image generator, and StabilityAI offers a paid service called DreamStudio.
In a statement, Stability AI said, “Anyone who believes this is not fair use does not understand this technology and has misunderstood the law.
In an interview with the Associated Press in December before the lawsuit was filed, Midjourney CEO David Holz described his image-making subscription service as “like a search engine” that pulls in a wide range of images from across the internet. I explained that it is. He compared copyright concerns regarding technology to how such laws have adapted to human creativity.
“Can a person look at other people’s pictures, learn from them, and make similar pictures?” said Holtz. “Obviously it’s allowed to people, otherwise it would destroy the entire professional art industry, and perhaps the non-professional industry as well. As long as AI is learning like humans, it’s kind of It’s the same thing, and if the images give different results, so be it.”
Copyright disputes mark the beginning of a backlash against an impressive new generation of tools (some introduced last year) that can generate new images, readable text and computer code on command.
It also raises broader concerns about the tendency of AI tools to amplify misinformation and cause other harm. For AI image generators, this includes creating non-consensual sexual images.
Some systems produce photorealistic images that cannot be tracked, making it difficult to tell the difference between real and AI. Most of the time, measures are in place to block offensive or harmful content, but experts say that it’s not enough and that people use these tools to spread disinformation and public I am concerned that it is only a matter of time before trust is further eroded.
“If you lose this ability to tell what’s real and what’s fake, everything will suddenly become fake,” said Wael Abd Almagheed, a professor of electrical and computer engineering at the University of Southern California.
As a test, AP sent a text prompt about Stable Diffusion featuring the keywords “Ukraine War” and “Getty Images.” The tool created photographic-like images of soldiers in combat, with their faces and hands distorted, with their guns pointed or carried. Some of the images also had a Getty watermark, but they were garbled.
AI may also indicate that they are not real, such as details on feet, fingers, or ears, but there is no set pattern to watch out for. You can also edit these visual cues. At Midjourney, for example, users often post in his Discord chat asking for advice on how to fix a distorted face or hand.
Some of the generated images can travel across social networks and go viral, but can be difficult to uncover as they cannot be traced back to any particular tool or data source. Washington using these tools for research.
“If you have enough experience with these tools, you can guess to some extent,” says Shah. “But more than that, there is no easy, scientific way to actually do this.”
But despite all the pushback, there are many people who embrace the new AI tools and the creativity they unleash. For example, searching for Midjourney reveals that curious users are using this tool as a hobby to create intricate landscapes, portraits, and art.
There’s plenty of room for fear, but ‘What else can we do about them?’ I did.
At the Museum of Modern Art, New York, Anadolu designed “Unsupervised” from artwork from the museum’s celebrated collection, including “The Starry Night,” and fed them into a large-scale digital installation to generate a mesmerizing animation of color and shape. To do. Lobby of the museum.
The installation “constantly transforms, evolves and dreams of the 138,000 old works of art in MoMA’s archives,” says Anadol. “From Van Gogh to Picasso to Kandinsky, an incredibly inspiring artist who defined and pioneered a range of techniques is present in this artwork, this AI dream world of his.”
Impressionist landscape paintings like Erin Hanson’s are so popular and so easy to find online that, for a painter who feels that influence in AI-generated visuals, she’s making her own $3 million a year. I’m not worried about prolific creations.
But she worries about the art community as a whole.
“Original artists need to be recognized or compensated in some way,” Hanson said. “That’s the copyright law. If an artist is not recognized, it will be difficult for them to make a living in the future.”
