How a DU Emergent Digital Practices Professor Has Embraced the ‘Wild Ride’ of AI Technology
Artificial Intelligence (AI) technology has accelerated rapidly in the last several years, expanding its possibilities for enhancing artistic expression while raising new ethical, legal and artistic questions, especially around producing artwork using AI generators. Colorado game designer Jason Allen recently filed a lawsuit challenging the U.S. Copyright Office’s rejection of an application for his Colorado State Fair award-winning piece “Theatre D’opera Spatial,” which used the AI software Midjourney.
The College of Arts, Humanities & Social Sciences (CAHSS) newsroom asked Chris Coleman, artist and professor of Emergent Digital Practices (EDP) and director of the Clinic for Open Source Arts to share his observations about the pros and cons of using AI technology, how EDP is paving a path to incorporate AI in the classroom and how the technology could shape the future of digital art.
Has AI technology changed the way you’ve taught your EDP classes and if so, can you provide some examples of that?
Well, we did call this department “Emergent Digital Practices" so, we have been following, testing and thinking about every technology that comes out, including AI, which certainly is moving a lot faster than some other technologies in terms of its most recent leaps. This kind of machine-learning technology has been out for five or six years, and we started teaching courses about it four years ago.
It’s the newer AI tools that have moved it from something hard to do to typing things into a website that gives you back art. It’s that consumer-oriented, super-easy part that’s new. We used to utilize this machine-learning technology by creating a database of images or other data — let’s say extinct birds — putting them into the machine-learning model and getting new birds that were imagined by the machine.
With some of these new systems like Stable Diffusion, they’ve basically trained it on every image from the last 30 years of the internet. The variety of things it can do are vastly different than training your own model with your own images and having control over that.
What about newer platforms that allow you to type in prompts for images and instruct them to create images “in the style of” a well-known artist?
I think that example of AI systems that name an artist is probably the most egregious and contentious for digital artists. If you want to sit there and type in something like “weird portrait of a man whose half of his face looks like a dog,” I think there is a case to be made for that being a valid art practice. But when you’re telling it to make it in the style of William Wegman, for example, that’s the tipping point that a lot of artists have problems with.
That would not be possible unless the images in those giant databases of images had been labeled with artists’ names. At some point somebody was also grabbing our works from places like DeviantArt [an online artists’ community] and others where artists put artwork up for free and their names were attached to it and that was fed into the database. If somebody had just labeled those images as something like “trippy tunnel with spacemen,” I think people would have understood that that was part of the collage but the moment you said “in the style of Chris Coleman,” for example, that opened the door to other people making art in my style.
Artists are suddenly reckoning with why they were posting their art online for free in all these places that don’t protect their copyright or personal livelihood. Once again, we have been reminded that, if something is free on the internet, you are the product — period. All these places just sold all the art you made over the last 10 years to some giant set of databases, and I think there’s going to be a reconciling of what it means to own your data and share art and how you want to participate in these systems.
And some artists don’t care. I try and make a lot of my art copyright-free or designate it as creative commons because I believe that all art feeds on other art. The question is what’s the difference between being inspired by and recapturing or re-constructing someone’s artistic sensibility? Artists have always had a hard time with that line between inspiration versus copycat and this has just made it crazy easy to cross that line.
What kinds of conversations are you having with students about how to approach AI art-generating websites and how to use them wisely to enhance artistic expression?
We have an intro to digital imagery class where you learn to edit images and collage multiple images together. So, we’ve opened that up to say, “instead of just taking your own pictures plus surfing the internet for stock photos, you can also try typing in prompts and getting new images from the AI machine.” But we’re really explicit that you can’t just submit that image, you need to process it just like anyone else’s photograph that you might use to make it your own.
Some of our professors are using some interesting AI technology. For example, Laleh Mehran is using AI tools in her video art class. AI has opened up some really interesting visual effects that used to take dozens of hours and require intensive labor that is typically offshored to places like India. Now you can use AI tools to do a lot of those things quicker and almost better.
So, it’s more about using AI tools to enhance the original art?
Exactly. In some of our cultural theory classes, we’re also reading texts generated by AI GPT chatbot-type systems and collaborating with them. Professor Trace Reddell and I have experimented with having students type things into the chat systems and having the computer respond to generate new conversations. And in a way, because the computer is using already written texts, you’re working on a collaborative collage of new and old knowledge and through the computer’s mistakes in language it creates new poetic conversations.
What about the lawsuit recently filed by Jason Allen? What are your thoughts on the issues the lawsuit raises and the broader conversation with students about artists incorporating AI technology in their work?
The U.S. Copyright Office [last month] released a new artificial intelligence initiative containing added registration guidance for the use of AI images. It still wouldn’t nullify this lawsuit, but they did have a blanket stance that you can’t copyright work produced by AI. Now it’s more along the lines that you can use AI, but it then needs to be significantly processed, retouched or manipulated to make it copyrightable. So, if I make a couple of AI-prompted images and then edit them and meld them together by hand, that’s copyrightable.
To be honest, we are going to have to accept and process a future in which it’s a valid thing to write these long complex prompts to create artwork. As far as EDP is concerned, I think that we would still hold to the principle that if you are going to name an artist in your prompt, you need to cite that and recognize the artist you are borrowing from. But if you’re not citing an artist specifically, there’s no way to give credit more broadly other than to say that was part of the medium for your tools. And that’s what I’m starting to see in art show submissions, the listing of AI as a medium used.
In EDP, we’re teaching our students to be aware of what they’re doing. I think in the long term these copyright guidelines will become null and void and we’re going to have to accept that [using AI tools and prompts] is just a different kind of artistic process. And yes, a lot of artists’ work got swallowed up and they’re going to be upset and make a big deal about it. But moving forward, artists need to be a lot more aware of what rights they give away when they’re posting their work on the internet. To me that’s the realistic future. I think [Allen] has every right to say this is my artwork, this is what an artistic process is now and these are the tools that are available to me.
Are students excited about using AI as a tool in their artwork?
The majority of our EDP and art students feel that anything that’s super popular online, from NFTs [non-fungible tokens] and crypto currency to AI, is dangerous, hyper-tech stuff. They’re turned off by the culture surrounding them. We have a couple of students who are all-in and want to use AI for everything, but not the majority. We had some conversations about it last quarter in my common curriculum class called “Understanding Digital Art” and there were students who believe using AI is theft and we spent a lot of time having a more nuanced conversation about it.
The issues around using AI in art projects are complicated but I’m glad we’re having this conversation and I’m glad to be teaching as part of a major that is exploring this space as it develops. I love that in our classes we’re not putting our heads down, we’re looking straight into the fire and that is fun and amazing but also a wild ride.
Listen to a related RadioEd podcast on the issues surrounding the use of AI technology here.
