At this year’s Emerald City Comic Con in Seattle, one of the panel discussions brought together several working artists to discuss how the advent of generative AI tools have already changed the face of their profession, the other problems they see going forward, and what they’d like to see from tech firms and legislators.
Actor Julie Snyder hosted a group discussion that included Fabrice Sapolsky, CEO of indie publisher Fair Square Comics; painter and illustrator Kit Steele; Tony Moy, watercolorist, writer, and creator of the WWII webcomic 4Forty2nd – The Lost Battalion on Webtoon; and Melissa Capriglione, a writer, illustrator, and creator of the webcomic Falconhyrste, who recently published her debut graphic novel Basil & Oregano at Dark Horse.
All four artists shared stories of issues they’ve had with AI art, both in general and with other creators, along with several proposed solutions.
“To me, you can use AI for translation, to write marketing copy, to check your grammar, to do a lot of other things that aren’t necessarily evil,” said Sapolsky. “The main problem is moral. We live in an era where technology goes faster than law.”
Sapolsky added, “AI is not going to go away. We have to get used to it. Now, how do we tame it?”
AI vs. reality
All four artists on the panel had run into issues that arose directly from the use of generative AI, particularly how it had affected their own practices and client interactions.
“One of the things that [AI] does, in terms of stealing, is eyeballs,” Moy said. “We all have a finite amount of time to scroll on our phones, and for every AI account that comes across your feed, that’s eyeballs away from someone’s actual art.”
An artist using traditional tools, such as pen and ink or watercolors, might take hours to create a single piece, whereas AI can generate dozens or hundreds of images in the same period of time. Someone like Moy might only be able to post new art once a day or less via social media, while accounts that are devoted to AI art can drown him out by sheer volume.
“I’ve seen Tony [painting] in hotel lobbies at 4 a.m. because he had a commission to finish,” Sapolsky said. “He spends eight, 10 hours on a piece, while someone can spend 25 seconds aping Tony’s style through Midjourney or Leonardo. That’s a problem.”
That in turn can and has set up unrealistic expectations from clients, who end up with inflated ideas about what a human artist is actually capable of doing.
“There’s an element of the level of expectation [AI sets] for human ability,” Steele said. “We’ve all seen the photos that show up on Facebook, like [sculptures] that just physically aren’t possible, regardless of the skill or talent of the original sculptor. You’re getting people who have completely handed their level of critical thinking over to the internet, and trusting what they see is reality.”
Capriglione added, “It feeds into the expectation on social media of instant gratification: ’This person isn’t posting enough, so I’ll follow this AI artist who’s posting more.’ We’re expected to post every day, but I want to work. I want to draw. C’mon.”
The result, particularly for a working artist that makes some amount of their overall income through commissions, is an audience with a loosening grasp on how, and why, art is made.
Art is not a democracy
A common argument by proponents of AI art is often that these programs “democratize” the ability to create, by removing the need for someone to train, practice, or get educated before they can work on some art. It’s a relatively common topic on pro-AI corners of social media, and Hopkins brought it up to the panel to get their reaction.
“Nowadays, people say that learning art is inaccessible because you need to go to art school,” Capriglione said. “You can go on YouTube, type ‘how to draw’ or ‘how to watercolor,’ and watch somebody’s process for free in 10 minutes. You can create with a $10 box of terrible markers. We can create with anything.”
Steele added, “What is there to democratize? Can you pick up a pencil? Great. You’ve got a start. Can you draw a circle? Awesome. You’re an artist. Keep going.”
Moy took a different approach. “Art was always about the singular perspective of the artist,” he said. “As a critique, as a commentary, on society, on how the artist views the world. Art is about gaining an interesting perspective from the artist’s point of view. To consider art to be ‘democratized’ also speaks to what we want the state of art and creativity to be, as [AI] tools become more prevalent.”
Rules of law
While no one on the panel argued for an end to generative AI, particularly when it’s used for personal entertainment purposes, a frequent topic was what could be done to legally protect artists in this environment.
“France has a rule that if you use Photoshop for any photography for advertising, you have to notify [viewers] that hey, this model was digitally touched up,” Moy said. “There are going to have to be laws and regulations that say, if it is AI, there’s a statement that says it was created with AI. We can create some simple rules and regulations to help us navigate the new environment.”
This could include methods by which artists can deliberately opt out or into their work being used by machine learning models, with a payment program for artists who choose to participate. It’s not simply the use of art as training data, but rather, that so much of it has been used without permission or even notification.
Sapolsky drew a parallel here between AI companies’ unregulated use of artists’ work and the American recording industry, which routinely tracks down unlicensed use of clients’ music on YouTube or Twitch. It’s a useful model for potential enforcement, should a company be found to have stolen artists’ work.
Offenders could readily be punished through methods such as algorithm disgorgement, where a company that’s found to be using training data without creators’ permission is forced to delete both that data and the products made with it. The FTC has already used or proposed algorithm disgorgement in several settlements, including a case last summer that involved Amazon’s Ring.
However, no one on the panel was particularly optimistic about that kind of pro-artist legislation being enacted in the U.S. The European Union has several initiatives and boards dedicated to arts and culture, which have no American parallel. That leaves American artists, for the time being, forced to organize and work on a grass-roots level.
This includes the ongoing lawsuit against Stable Diffusion, which has several Pacific Northwest artists such as Sarah Anderson and Phil Foglio attached as plaintiffs.
At the end of the hour, the general impression I walked away with was that the issue here isn’t the existence of generative AI tools by themselves, but instead, how they’re used and often misused. This was particularly driven home by Sapolsky, who frequently spoke up to point out that AI in general is simply a machine. It’s not good or bad by itself.
However, it does represent a potential branch point in arts and culture, particularly in our relationship with what and how we choose to create.
“What do we want art to represent?” Moy asked at the panel. “Safe stuff? Clean stuff? Does it challenge our thoughts, how we think and how we perceive? That’s a question that we all have to think about and answer.”
