AI Art: Promise and Peril
A discussion last week at Syracuse University
I had an interesting discussion a few days ago about AI with the art students at Syracuse University. I wasn't even going to bring the subject up. The subject of AI art felt to me like a toxic swamp filled with wounded birds.
But my host, Illustration Department Chairman Frank Cammuso, said the students wanted to talk about it. So we did.
We discussed both the threats—and opportunities—of generative AI and machine learning as it related to art. In this post I’d like to share with you my preliminary (and very mixed) musings and invite you into the discussion in the comment section.
Prologue: Animating a Wave
I started off telling them about my crazy Twitter post a couple months ago. Maybe you saw it:
As you might expect, the response to that query on Twitter ranged from angry to disappointed to intrigued. After the shouting died down, an artist who is also a computer expert privately sent me the following video clip that he generaated using the Hunyuan video model:
It’s not perfect. It goes in and out of focus, it looks a bit too photographic, and it’s not the whole cycle. But it gives a hint of how AI can bring drawings to life.
The person who prompted the AI animation said to me:
“My actual career is in IT, and I was a fine arts major in college. I will never give up traditional media but I also appreciate the tremendous possibilities to explore different perspectives and ideas in visual art. I liken it to the camera's effect on the Impressionists.
“I also appreciate the angry response from people, but I also see the possibility.
“I tell my kids you have to get in behind this AI revolution if anything to keep tabs on what is possible: to use and understand these tools but to not abuse them and continue to work without its help.
“It is the future for sure and us old artists need to know this stuff.”
The argument about copyright scraping
Many artists focus on the problem of how generative AI models use copyrighted works for training without permission, credit, or payment.
I understand those concerns. But the argument has problems, and it’s a slippery legal position.
Lets start with the Ghibli craze a couple weeks ago. Should Studio Ghibli be able to stop ChatGPT users from generating images in their style? They’ve been silent in the wake of the recent meme-storm, but founder Hayao Miyazaki was very clear in earlier comments, where he said this technology is "an insult to life itself".
Broadly speaking, should an artist have the right to own and control their style? Should they exclusively own the ideas they innovate, or should they own just the particular expression of their ideas? Does it matter whether a human or a computer uses the inspiration? These are important distinctions.
Or consider, for example, Steven Spielberg. Should he be the only person allowed to make movies about visitations from friendly aliens?
I’m not convinced that Studio Ghibli is harmed by the attention or that they have been diminished in some way. If human artists — or humans using AI tools — want to invoke his style, it doesn’t really take anything away from him. In fact the opposite might be true. It’s better to be copied than forgotten. The worst fate for any artist is oblivion.
To my mind, copyright-scraping is a small part of the overall picture. There are other issues that are, I believe, more fundamental and far reaching.
Artist and audience
Many people assume that artists and followers will remain defined and distributed more or less as they are now, but that's one of the things that's about to fundamentally change.
AI will alter the relationship between artists and audience. Skilled creators may or may not use these tools, but either way the audience may feel empowered by them. There will always be passive fans, but there will be a new class of artists that we might call "AI-enabled sub-creatives."
Smart world-builders will enlist the help of a team of these creative allies, who can help them not only create the world in all its detail, but also be a liaison to the majority of people who experience works more passively. For decades there have been fanzines and cosplayers, but that dynamic of the empowered fan is about to explode.
Psychological effects
As AI increases in capacity, what happens to us psychologically? If we allow AI to usurp the human acquisition of skills, how does a person feel who is proud of their hard-won knowledge?
What does it do to our mental health to have a creative partner that comes up with good results so easily? What does it do to our confidence and drive?
I once knew a person who was married to a really smart and talkative spouse. That spouse would always answer first. This friend of mine eventually gave up and let the spouse answer everything.
Computers generate solutions coolly, quickly, efficiently, and politely. Humans, on the other hand, create things slowly. Often good ideas bubble up from a stew of emotional turmoil.
What are the implications of an AI model that refuses to cooperate? What happens when it tells me that I’m not supposed to imagine something? What if it becomes illegal to imagine something without using a safe-certified AI?
Corporations will inevitably erect more content guardrails. Governments will use AI more for surveillance or mind control. If that day comes (some say it’s already here) we may find ourselves trying to make sense of the world while those in power are waging a war on reality.
We haven’t fully seen AI’s more ominous face yet. What might happen and how people might react is impossible to predict and it’s the stuff of science fiction.
Staying afloat in a sea of slop
Those are dark thoughts. Let’s step back and ask ourselves: Is the world worse off or better off now that we have released this art-making genie from its bottle?
It’s hard to say because everything is changing so fast. We’re already drowning in AI images. It’s getting harder and harder to find human-created art. For consumers of art it will become increasingly difficult to tell how or whether they used AI. More and more creatives will use AI for different parts of their process, but you'll never know it because they've gone underground to avoid ridicule or doxing.
Among the leading artists, even those whose output is mainly analog, it will be a challenge to navigate the rising sea of slop, especially if you’re working primarily with digital media. For those trying to detect the use of AI there will be lots of false positives—accusations of using AI when the artist really didn’t.
The rise of the renunciates
Old-school artists shouldn’t be snobby toward their digital siblings. It’s all art if it moves us. We’re all artists. No one is guarding the gates, but it’s a lot more crowded in here.
But there will be some people who can’t stand art made with AI. Those people will want their art to be certified “FFTM” (Free From The Machine).
Unfortunately "AI-Free" can only be proven by filming things as they are made by hand, or better yet, watching someone actually demoing a painting or playing an instrument.
People like Mark Zuckerberg probably underestimate how many people will resist or reject AI whenever its use becomes obvious. Some will renounce digital life altogether and stay as far from their computers and phones as possible. This movement doesn’t yet have a name, but these people may call themselves something like the Humanists or the Renunciates.
To satisfy that expectation, creators will have to be forthcoming when asked about their process. The only thing that is dishonest is lying about how you created something.
What sets apart the leading artists?
Whichever way the work is created, a lasting creation should be more than just attention-getting or meme-worthy. It has to capture the public’s imagination, move the emotions, and reflect the zeitgeist.
Whether artists use AI or not, they will be known for their taste, knowledge and authenticity. What will set creatives apart will be an awareness of the history of art and how they dance with that history. To be a leader in AI, you’ve got to have all that contextual knowledge, but you also have to understand and manage the ever-growing range of models and tools.
New art forms will be created by combining AI with analog skills—old school, handmade, analog techniques. That blending of old tech and new tech is similar to what Laika did by combining stop motion with 3D printing, or what the Cuphead game designers did when they used drawn-on-paper animation to guide the look of a modern video game.
My recommendations to young (and old) artists
I agree with the Artist/IT person who sent me that wave animation. It’s a good idea “to keep tabs on what is possible.” I recommend that people go ahead and play with whatever tools they think might help them solve a problem.
For now, it might be wise not to put your flag in the sand and declare absolutely what AI tools you will use or won't use. And please don't scold or shame people who use AI. I would rather live in a world where people can openly share how they made things.
Each of us has to make our own peace with the digital world. The technology is changing so rapidly that if you want to keep up with the latest tricks and tools, you’ve got to be willing to throw aside all the tools you used last week. Or, as most of us will end up doing, you’ll patch together a personal workflow that gets the job done for you using a mix of new and old tools. But unless we have to conform to a standard company workflow, everyone will come up with their own mix of old-school and new-fangled methods.
One thing I’m sure of. The most immediate outcome of the advent of genAI and machine learning is to put a value on live performances, acoustic music, and painting workshops. If it drives a lot of artists back to the basics of sketching and journaling, maybe that’s a good thing.
Resources:
AI-curious YouTube channels: Matt Wolfe, Theoretically Media
AI-skeptical channels: Gary Marcus, Ed Zitron
Please leave a comment. I’d love to know what you’re thinking about this.
I think this is a really thoughtful discussion and I’m glad everyone is contributing interesting view points. However, I think that speaking about AI art the way we would talk about Photoshop or Illustrator is a mistake. This is not just a tool fueled by Stock photos that were purchased/licensed, or an algorithm that creates interesting digital brushes. The only way that AI art can exist is by stealing a vast amount of work from other artists, the majority without their consent, and training the algorithm based on the stolen artwork to create more artwork. It’s a much bigger problem and conversation than, is copying a style plagiarism? Because the AI art CANNOT exist without the artwork that was stolen. That’s why I think it’s a more serious conversation than, what a fun new tool? How can we use it ethically? Because its formation and existence is unethical.
3 points to add which are often missed:
1. It is not a tool. It's a service. The corporations control access and possible content creation. No pencil watches what you draw. Midjourney has a list of forbidden keywords. Which ones? It won't tell you. People who call themselves prompt engineers pay for being employees helping to refine the models.
2. The LLMs are a byproduct of a larger goal which will in the end help to make totalitarian visions become a reality. Along the way, it disrupts markets with potentially billions of people, washing the accumulated money into the hands of a few owners who possibly had the training phase of their models funded publicly. We pay the autocrats in order to help hollow out democracy.
3. Artists and activists are fighting back on the same battleground. Technology like Glaze for visual artists actively attack LLMs at their core. Benn Jordan allegedly is helping to do the same for music, right now. That, in the end, is our only hope (Obi Wan Kenobi), as legislation is too slow, public discourse intentionally made toxic and courts busy trying to understand what is going on.