Social media sites have long been a place for artists to share their work and creative vision. But now, one might have to look twice at the images they see on their Twitter and Instagram feeds.
Sometimes, these images might look like a digital painting of a vibrant landscape, and other times, it might be an illustration of Walter White from “Breaking Bad” in a Starbucks barista apron. The thing these images have in common is their origin – they were created not by brush or pen, but through code.
Artificial intelligence-generated art has seen a rise in popularity across social media platforms over the past year as computer programs like DALL-E, Midjourney and Stable Diffusion are being used to generate these pictures.
“AI art refers to programs that analyze huge data sets of images and text that is attached to each image so that we can use our own text to generate brand new images,” said Mats Borges of USC’s Ahmanson Lab, an extension of the Sidney Harman Academy focused on interdisciplinary research and experimentation.
In a nutshell, an individual writes a prompt and the program looks at thousands of images across the internet with similar words attached to them to create something new. For many, this is an exciting breakthrough that bridges the gap between art and technology, but a debate has slowly come to the forefront regarding the use of AI art.
While it started as a fun way to generate bizarre images, some artists are voicing their concerns about the technology, with some even calling AI artists “nightmare clients” when attempting to pass their work off as “real” art.
This heat comes off the tails of an AI artist winning a fine arts competition. While the judges let him keep the money, the incident was still met with backlash. Now, a steady rise in AI-generated images have begun to circulate the internet, with some defending them and others criticizing their classification.
The lingering question now is whether AI art is “real” art.
Of course, the first question to raise is the definition of art itself. To some, it is a form of expression, and to others it is simply a way of bringing human creativity to life. Elea Zhong, a sophomore at USC’s Harman Academy believes that art can be split into two categories.
“For some people, they think that art is solely used for artists to express their ideas and emotions and passions and politics and culture,” Zhong said. “But I would also argue that at the same time, besides that, there is art that isn’t necessarily emotional or passionate, like corporate art or the art on billboards or Coca-Cola bottles.”
However, to others, art also includes the human technique and practice put into artistic creation. They feel that carefully crafting their sketches or digital drawings cannot compare to an individual sitting in front of a computer and typing a few words into a machine. Borges explained that this debate has happened before with something that is now recognized as a hallmark in the art community.
“I think the best parallel, and you’ll probably start hearing more of this too, is comparing it to when photography was first invented,” Borges said. “They were like, ‘This is going to put all the artists out of business, there’s going to be no more portraiture,’ [and] that it’s all going to be photos. ‘Photography is not real art, you just click a button and that’s it,’ and that’s not really true. I think over time we have all generally as a society come to accept photography as an art form.”
On top of this, there is also the idea that many consumers purchase art not for any expression or technique, but for aesthetics. According to Zhong, many people think little of how pieces come to be, and pay closer attention to how they look. She said that from “live, laugh, love” signs to the decorations people hang in their dorms, there are several pieces of art where people are less interested in their creation and subtext.
“We’re not really thinking about emotions and the cultural background and history. We’re just seeing that it’s a pretty thing that is in our rooms or it’s a pretty thing that’s my background wallpaper, or it’s a pretty thing that’s on a sticker. Not necessarily like an emotional piece,” Zhong said.
Even with these qualifications, there are ethical concerns regarding AI-generated art in regard to the types of images it uses and the styles it borrows. Borges gave the example of how typing in the word “CEO” will most oftentimes generate images resembling old, white men given the datasets they sort through.
This is a fault that algorithms across the internet, even in search engines, have had since the advent of this technology. As author Safiya Noble once discussed in a 2014 TEDx Talk, the top Google auto-fills for “why are Black people so…” included words like “loud,” “athletic” and “lazy.” While these algorithms have evolved over the years, many of these biases persist and bleed into the programs generating AI art.
“What’s in the data set? How are they combing through things? What do they choose to say, ‘this should be in the data set?’” said Borges. “That’s not even talking about artists whose whole careers have been building a style, especially concept artists. And now, someone can just type what they want and say in the style of blah blah blah and take that while the artist never consented to having their [art] in the dataset.”
In a way, these machines could be seen both as biased and as thieves. These flaws ultimately exist because of the lack of direct human involvement in their creation.
“It has no comprehension of what these images are…Type in paintings of bodies, like artistic paintings, nude images, etc. and what comes out is really grotesque,” said Borges. “You will get like these double bodies where one person’s leg becomes the neck of another person and like their arm becomes a foot and their foot is a hand.”
This returns to the question: can this be considered art? On one hand, in aesthetic terms and historical precedent, it perhaps could be put under that umbrella. But these machines still have holes that make them incomparable to the standards of human creation and leave them vulnerable to abuse.
Regardless of the answer, it seems that AI art will remain in this increasingly digital world. For Zhong, there is immeasurable potential for its continued growth both technically and socially.
“I feel like [normal artists’ spaces] are valid in disliking it and I feel like there has to be some way that we could try to get them to accept it or try to create tools that are not intrusive and not abrasive and not like touting its revolutionary, etc. that would actually help normal people to create art,” said Zhong.