Indiana University students and faculty are using AI to make important advances across multiple disciplines. Researchers at the Luddy Artificial Intelligence Center, for example, investigate the challenges and opportunities of AI from technical, societal, scientific and applied perspectives. Their work is making a difference with a variety of projects, including the development of robots that assist older adults, improvements in speech recognition and technology that automatically flags cyberbullying.
In the IU Eskenazi School of Art, Architecture + Design, associate professor of painting Caleb Weintraub and lecturer in photography David Ondrik developed an interdisciplinary course called “AI in the Studio,” which was taught during the recent spring semester. The course explored how text-to-image AI generators can be used as a tool in the creation of physical art as opposed to being used to replace artists. They also studied legal and ethical issues of AI-generated art.
Inspiration for the class came after Weintraub found AI to be a useful tool in his artistic process. He first experimented with rudimentary AI art programs during the early stages of the pandemic, asking a program to interpret one of his paintings.
“I characterized the painting as appearing somewhat fragmented and difficult to read but mentioned that it depicted a figure on skis slogging through a snowstorm,” Weintraub said. “When analyzing that painting, the AI explained that the artist had likely chosen to make the image difficult to read to give the viewer a sense of the skier’s experience of being disoriented by the storm.”
What the AI did next delighted Weintraub.
“The language model proceeded to divulge metaphors, critique palette choices and provide uncannily subtle and incisive interpretations of potential ways of understanding the paintings. I was immediately struck by the promise of AI as a tool in the studio — a critic and collaborator.”
Weintraub said he put his new copilot to work when commissioned to create three-dimensional, articulated sculptures of characters from the classic film “A Streetcar Named Desire” for the stage set of John Mellencamp’s 2023 tour. Blurry film footage and shadowy, two-dimensional images from the 1951 film left him little to work with as he re-created the likeness of stars like Marlon Brando and Vivien Leigh.
“By using an AI upscaling tool, I was able to enhance and clarify the blurry and indistinct images of the actors,” Weintraub said. “This considerably improved the detail in the images.”
The good, the bad and the unknown
Concerns surrounding AI have long been depicted in fictional stories, with the earliest literary references to artificial intelligence found in Samuel Butler’s 1872 novel “Erewhon.” From the benevolent droids R2-D2 and C-3PO in “Star Wars,” to the self-aware AI system Skynet that attempts to destroy mankind in “The Terminator” movies, to the lifelike title character of “M3gan,” the positives and negatives of AI continue to play out in art and popular culture.
Whether AI eventually creates a sentient life form that takes over the world or not, the current technology has artists and storytellers concerned about the future of their occupations. The Writers Guild of America is still on strike, demanding writers’ work be protected from being undermined by AI in the future. Weintraub and Ondrik addressed both concerns and positive uses of AI in the classroom.
Students enrolled in the new course had mixed views of AI art. Avrie Allen, a rising IU Bloomington senior studying graphic design at the Eskenazi School, said she went into the class with a relatively anti-AI perspective. Despite her apprehension, she and other students tried to have an open mind.
“I think that there was a good mix,” Allen said. “Some students were very into transhumanism, versus people who said, ‘Do you have no respect for yourself as an artist using AI?’ It was one extreme to the other in our class. I think most people used it earnestly and explored it as an artistic tool.”
New technologies have historically been met with distrust, and Ondrik likened the current chatter about AI art to what painting went through in the mid-1800s as photography emerged.
“When photography was invented, a very precise, optically accurate landscape painting became less interesting because the photograph can do it so much faster,” he said. “What’s interesting is you’d think portrait painting would have been the first medium to go, but the historical record indicates that that is not true. People who couldn’t afford to have their portrait painted were able to make photographs, and people who could afford to have their portrait painted still had their portrait painted. It democratized portraiture by allowing more access to it, just in a different medium.”
Weintraub said he believes AI will make creating art more accessible.
“Now everyone has access to take the imagery that is inside their minds and bring it outside and contribute to the visual culture,” he said. “I think that’s a positive side because it creates more access to more people.”
The “AI in the Studio” course used the technology in various ways. They implemented AI language tools to brainstorm ideas for physical art pieces and asked the program to interpret artwork, using it as a pseudo-critic. The class also experimented with image generators by using a work of art they created and asking AI to make varied iterations of the same piece.
Weintraub compared prompting an AI art generator to create an image from one’s mind to “casting a spell.” He said one of the most interesting and positive discoveries he and Ondrik made during the class was detecting individual students’ aesthetics through the images they prompted.
“We’d look at something on the wall during a critique, and David and I would often know who the creator was because we could see characteristics of their taste,” Weintraub said. “I think that everybody who has a bit of a style, or an inclination visually, is able to bend their spells in different directions.”
Harnessing the power of language is integral to prompting an AI art engine, and Weintraub said it’s a skill that must be honed over time. It takes time to learn which prompts will work and which will not, and he realized that prompting the AI engine was making him a more streamlined communicator.
“In terms of the way I try to communicate an idea, I often now will think ‘Oh, how might this be misinterpreted if I don’t qualify it in a certain way?’” he said. “It’s almost like speaking to somebody who’s exceptionally good at language, but English is their second language.”
Allen said she appreciated the course because she learned a lot about the technology. In her final project, she explored concerns of privacy and the labor conditions that go into creating AI models. Another concern she mentioned was the energy required to train AI systems, comparable to mining Bitcoin, which consumes a lot of energy.
“I don’t anticipate using the AI programs we utilized in class because of my concerns and their lack of relevance to my work,” Allen said. “However, there is more critical engagement with AI on a smaller scale than these larger companies, like artist James Bridle’s work, which I find exciting.”
The ‘Wild West’ of AI
CNN’s Chris Wallace recently asked actor Harrison Ford about the use of AI to create a decades-younger version of himself for the film “Indiana Jones and the Dial of Destiny.”
“I think it’s not a question of the technology; it’s how you use it,” Ford said, placing the onus on practitioners rather than the technology itself.
The technology is currently in its “Wild West” stage as legal and ethical issues come to light. Among them are the ease with which AI can create deep fakes, AI-generated scams, misinformation and propaganda, as well as ethics in the development of AI programming. The legalities of copyright are a particular concern for artists and creators.
Weintraub and Ondrik addressed such concerns in the IU course, in addition to teaching students to experiment with AI. They invited Nazareth Pantaloni III, copyright program librarian at IU Libraries, to speak to students about the legal issues surrounding AI art.
“Under U.S. copyright law, copyright protection does not extend to non-human creators, which is an obvious problem for AI-generated works,” Pantaloni said. “The Copyright Office will not register works produced by a machine or mechanical process unless there is some copyrightable input or intervention by a human creator.”
The issue has already made its way to federal court.
“In a case that is currently being litigated in a federal trial court (Thaler v. Perlmutter), the inventor of an AI image generator sued the U,S, Copyright Office for its refusal to register a work created exclusively by AI,” Pantaloni said. “The federal District Court for the District of Columbia has not yet announced their decision, but it is likely that they will uphold the longstanding legal principle that only works by human creators are copyrightable.”
According to Pantaloni, the legal issues pertaining to AI will take years to resolve. The U.S. Copyright Office is expected to announce a call for public comment on copyright and AI by the end of 2023.
It is yet to be determined if “AI in the Studio” will be offered as a course in forthcoming semesters. However, a bevy of courses related to AI are available to prepare IU students for the future, including as it relates to cognitive science, business, law, linguistics, education, informatics, economics, politics, philosophy, information warfare and cybersecurity.