Artificial intelligence could be a new artistic tool, or artistic challenge
While people may notice when images are completely generated by artificial intelligence, they may not be able to identify when a photograph or illustration has been enhanced by its use.
Katherine Mickle, associate professor of art at SRU, said her students use AI “minimally to add missing information into their photography or other digital work.” The technology can help photographers clean up digital images, but artificial intelligence programs can’t make up for lack of skill in framing or shot composition, she said.
“I had a student last semester who showed me where they got rid of a visually interfering subject in the corner and used generative fill to input the side of the furniture that had been missing from the removed subject,” Mickle said. “Had he not shown me how well it worked, I wouldn't have known it was AI.”
Students in assistant professor of art Doris Short’s graphic design class are working on a project for the university’s president, Karen Riley, but have concerns for their future in the field where the potential of art generated by artificial intelligence is growing.
Andrea Phelps, a senior graphic design major at SRU, said Thursday, Feb. 8, she isn’t immediately worried about her career prospects in graphic design, but has some concerns about other applications of images made by artificial intelligence.
“When you type in a generative caption, and it creates an entire image, it’s worrisome because not only does it limit creativity, but it also discredits artists working decades to get where they are,” Phelps said. “Companies will then start to use it as a normal form, so that they don’t actually have to have a hired person for that.
“They can just generate an entire marketing campaign at the touch of their hands.”
Sean Macmillan, chairman of the art department at SRU, said AI programs that generate images are creating pictures based on other images that already exist on the internet. They take a sample of images and “regurgitate” an image based on related images.
“AI basically regurgitates everything that is already out there,” Macmillan said. “It's not going to come up with something completely on its own like a human can.”
Macmillan teaches metalsmithing at SRU, a program that also teaches students how to create designs to be molded by a 3D printer.
“(Metalsmithing) was one of the first spaces to embrace 3D printing,” Macmillan said. “It hasn't put them out of business; it's just a new tool we can utilize.”
Adobe programs used by students and staff at SRU come loaded with features that make certain tasks easier for their users. Adobe Photoshop and Adobe Illustrator have become more and more advanced over the years, Phelps said.
Jon Shumway, professor of art at SRU, said he has had a few students utilize aspects of Photoshop's AI generative fill to assist with the development of compositions. He said this has mainly helped make designing more efficient.
“These are new tools and, in the hands of artists, they will serve to expand the possibilities of what artists can create and accomplish,” Shumway said. “These tools will, in all likelihood, alter or expand how we broadly consider what we do as artists.
“But then, the only constant in this world is change, and it is coming.”
Kalen Johnson, a senior graphic design major at SRU, said some digital programs allow users to manipulate photos by removing shadows or clouds, but it is still difficult to make it work, and harder to make it work convincingly. Manipulating photos digitally still takes human effort, she commented.
“It’s very hard. I had to type a couple of times what I mean to get it,” Johnson said. “It’s very new; it’s very touchy.”
Johnson added artificial intelligence’s inability to create something brand new could hurt the originality of art in the future, even when it comes to subjects like company logos.
“When a brand comes to you, and they want a logo done, you look at your competitors and what they’re doing and you want it to be unique, you want it to be different,” Johnson said. “AI is just using data that is already there ... so is it really unique? How long can you go until you need another artist to make a new idea?”
Some artists have used images generated by artificial intelligence as reference to draw or paint their own picture, according to Phelps. This can be a helpful way for someone to create their own art, but Mickle again said it depends on an artist’s use of it.
“Like many shortcuts in any medium, is it convincing? What does it convey? Does it do what the artist intends?” Mickle said. “It's here and not going anywhere, so we just navigate those waters, especially as it becomes more advanced.”
Macmillan said the art department hasn’t fully integrated AI into lessons for students, but AI images are becoming more prominent on the internet, and even in certain art shows and galleries. He said people are already entering generated art into shows, and the submitters are open about it.
“I have juried high school art shows, and there is an AI category in the scholastic arts,” Macmillan said. “No one is claiming to have made it on their own, they just give it the prompts.”
Short said despite concerns she has about the role of artificial intelligence in design, learning about its application could be useful for students coming through art programs at the university.
“I think that AI has the potential to revolutionize the way we approach design and lead to better, more efficient, and more innovative solutions,” Short said. “There are opportunities in this new space and my students should take advantage of what AI tools offer.”
Abion Levert, a senior graphic design major at SRU, said digital programs have become better and better for use improving work done by artists like himself.
“It’s surprisingly good at recoloring your work,” Levert said. “It’s not very good at making images from scratch, but it is good at changing the colors.”
Matthew Jumba, a senior graphic design major at SRU, said AI generated images that are used professionally typically involve some human input, from giving the initial prompt to reviewing the final product.
“It’s not going to replace a person,” he said. “You still need someone to look at what it has generated and do something with it. We’re getting in pretty early with it.”
Phelps said her concerns with the advancement of image-generating technology lie more with its use to spread misinformation. Having already been temporarily convinced that a generated image was real, Phelps said, it will only become more difficult.
“It’s not that I’m worried it’s going to take over,” she said. “I’m just worried that people are going to take advantage of it and create mass hysteria. Because they look so real, it could be dangerous.
“Typically people skip over it so fast that our brains don’t compute that it’s actually not real.”
