In the past, a masterwork of art might have been painstakingly copied for weeks to be sold to an unwitting art lover or collector. Today, with AI technology, the work of just about any great artist in human history can be mimicked in moments to produce an image on spec by a user — and openly sold.
AI generators rely on databases of already existing art, the internet being a massive trove of readily available resources. Images are “scraped” from the internet and used as data. Generators are “trained,” or taught, to recognize art. Based on human prompts and drawing from thousands or millions of images, They can produce a new image that meets particular specifications.
Many images used as the basis of replication are in the public domain, but artists charge that copyrighted images are used as well, raising questions about the uncompensated use of the intellectual property of others. Artists who make their living by their work may go uncompensated for the use of their creations, or even find themselves priced out of some markets by less expensive AI-generated images.
“If millions of artists’ works are being unknowingly and/or unwantedly used to enable (to train) AI systems,” says Selwa Sweidan, clinical assistant professor of multimedia arts in the LMU College of Communication and Fine Arts, “so far it seems to happen wholly without the artists’ consent and without their benefit.”
Sweidan points out that AI platforms enjoy an asymmetry of power in the form of economic resources that individuals and few entities can make use of and a high barrier to entry that limits competitors. But redress is possible, even it appears to be a game of catch-up. “Within a capitalist system like ours, legislation and oversight are ways to attempt to reign in the harms that happen to the less advantaged and the marginalized.”
Another form of redress is to use technology to fight technology.
Ben Zhao, professor of Computer Science at the University of Chicago, has worked with his colleague Heather Zhen to develop a program, called GLAZE, that will protect artists’ work from unauthorized mimicry through the use of platforms such as Stable Diffusion, Midjourney and DALL-E.
In a keynote address on art and AI at a March 2023 conference in Chicago, Zhao pointed out that artists may not only face a loss of income, they may find their artistic identity — their brand — diluted or, as some say, polluted. A knock-off of a Picasso may satisfy every wish of a user, and, no doubt, will be cheaper. But because users of AI platforms can prompt systems to generate specific images in the style of a particular artist, any well-known and highly respected painter, illustrator or photographer working today faces the same prospect.
Zhao said some artists may stop promoting their art on their personal website, to prevent their work from being scraped without permission. Some, he added, are experiencing signs of stress and depression, and young artists are considering whether to leave their field. “AI art models can’t yet replace human artists via ‘training,’” he said, “but [AI] is already disrupting art industries.”
Zhao pointed to the example of Karla Ortiz, a San Francisco-based illustrator and fine artist. She has worked on films including has seen her style of “Avengers: Infinity War,” “Black Panther,” and “Rogue One: A Star Wars Story” and “Doctor Strange.” She has seen her work mimicked with AI technology to create works that resemble her well-known, highly valued style. In order to protect the work of Ortiz and other artists, Zhao has created GLAZE, a tool that allows an artist to post their work online in a way that when scraped generates an altered rendition of the work, disrupting AI-based imitations. A painting originally rendered in an impressionist style, for example, may appear in an AI-generated form in an entirely different style — abstract, for example — and thus useless to those hoping to mimic the work of artists.
GLAZE, then, is a tool in the battle against copying. Yet, copying has long existed in the art world, and often in a useful way.
Damon Willick is a professor of art history in the LMU College of Communication and Fine Arts. In the past, he says, artists in training copied the work of masters in order to learn the craft and develop their own aesthetic. More recently, artists like Andy Warhol or Marcel Duchamp co-opted existing imagery to create new images that critique works of the past.
“AI isn’t doing either of those things,” he says. “It’s doing its own copying. It doesn’t strengthen the image that it copies, neither does it critique that image.”
The question of the relationship of reproduced art to the original is a central concern of José Garcia Moreno, professor of animation in the LMU School of Film and Television and director of the Academy of Catholic Thought and Imagination.
Although he is concerned about theft with the tools of artificial intelligence, Garcia Moreno also sees creative possibilities, especially in terms of expanding human understanding of the world in which we live.
“If we only think about using artificial intelligence to create machines that will think for us instead of helping us how to reason the world, that is a problem,” Garcia Moreno explains. “But if we use machines to help us to reason the world because we’re creating a language and interacting with a language, then it’s only expanding our consciousness.”
Garcia Moreno believes artificial intelligence will create an even deeper desire for audiences to go beyond reproduced art and seek out the real thing. “The physicality of art will be very much desired, because it’s the real thing,” he says. “We will go back to understanding what is real and what has been reproduced.” The reproduced image, in other words, will highlight the beauty of the original.
Everything will be in the public domain eventually, Garcia Moreno says. “When things enter the public domain, everyone can use it. The old Mickey Mouse from the 1928 Steamboat Willie film is going to appear in many of other forms because now it is not restricted. Everything is fair game at some point. What is fair game now for artificial intelligence? Of course, we want it to be conscientious of the value of original work.”
Garcia Moreno is working with new media by using 2D digital animation that uses a mobile device app to overlay animated images found in the St. John’s Bible. That bible is a project of Donald Jackson, artistic director and principal illuminator of the St. John’s Bible, a handwritten and illuminated bible.
He hopes that his app will draw young people to an appreciation of Jackson’s work. Garcia Moreno sees his project as adding to pre-existing art, which, of course, stands within the long tradition of creating contemporary art speaking to speak to earlier art. His work, he says, “has a purpose, a goal, it’s done with a lot of respect. It’s just an exploration of what has been done, to create a new audience.” Nonetheless, Garcia Moreno believes that ethical frame is essential.
Garcia-Moreno grants that misuse of capabilities will occur, but the future lies in not stopping the evolution of artificial intelligence but, rather, in ensuring that the creative possibilities of artificial intelligence are not channeled in a harmful way.
Sweidan, while deeply concerned about issues such as who benefits from new technology, is reluctant to judge a new technology as either good or bad. Photography was once considered by some, she points out, to be a threat to non-photographic art. “I don’t think any contemporary artists would consider that position for a moment at all in 2023. … For me it’s more about thinking about how a system is made and how do we allow it to be used.”
Nonetheless, Willick believes a strong dose of skepticism is called for. He considers AI-generated art “flat and very cold,” precisely because it lacks the crucial human creative element.
“I don’t think there is any stopping this technology,” Willick says. “What I hope happens is that there emerges an avant-garde that pushes back and that creates an alternative that’s more human, expressive and emotional, so that people move away from the allure of the easiness of AI-generated art.”
Joseph Wakelee-Lynch is editor of LMU Magazine.