Transcript:
This interview has been edited for length and clarity.
Luis Hernandez: How are you teaching AI in an art class, in an art school?
Griffin Smith: We have a once a week studio course from 1 to 6 p.m. And that’s the same model, if you were an oil painter or a glass blower and in a traditional setting, you fill up all that time with the materials, you let the glass cool, you make things for hours and hours with a studio class, with these kinds of tools. You’re really retrofitting it for what feel like brand new questions. You know, how is it going to be my art if I make a thousand of these in a day. Photographers ask that question. Printmakers and ceramics artists ask that question. There’s a rich history of how artists deal with automation or transformation like digital tools.
So one thing that I think is a real gift of the tech is when my students ask questions that they think are very new within the past two or three years. We can couch those questions in art history. We can compare it to what the camera did to the photography majors on campus. We can compare it to how the internet disrupted art sort of in general. So even though it’s a studio course, we make stuff, and the questions that we ask end up sort of being art historical questions.
Hernandez: It seems to me [that] out in the world, a big question people have is, “How is AI going to replace me?” Do students ever ask, “Wait a minute, how is AI going to replace me as an artist?”
Smith: Absolutely. We have senior students who are looking at the job market and saying, “Every job that I am applying to right now has AI in the job description and says we want you to know how to use it.” So we see it day-to-day adjusting for the seniors. And then you have first years at RISD who say, “What is the job market or the art market going to look like in four years’ time when I graduate?” And to have both of them in a class together sharing their concerns, it does really put things into context.
Hernandez: I remember hearing from you the last time we talked, you told me that artists aren’t as worried as designers are worried.
Smith: Yeah, there’s a sort of tension on campus because we’re an art and design school, and in some ways artists and designers don’t really have that much in common. An artist, like an oil painting major at RISD, they’re going to be hired because they have a weird point of view, they have an aesthetic that’s all their own, and when culture changes, you know, if there’s a war that breaks out or COVID happens, the fine artists are there to critique it and to make weird stuff and to sort of represent this other voice in culture. And so the artists say, “If AI represents a crisis, I’ll critique it, I’ll make a great oil painting about the machine, and I can sort of do my role in culture. I can critique it as an artist.” My designers say, “Well, all day long, I get stuff from a client who says, turn this into an image, or turn this into a different kind of language.” And they see themselves doing what these AI models do all day long. They get in a text prompt, whether it’s an email or a brief from a client, and they say, “Make this into art.” And that seems like what the AI models do. So it’s a very different question for my designers when they say, “How am I going to be replaced in a practical, professional workflow?” My oil painters aren’t concerned about that.
Hernandez: Tell me one thing that excites you about AI in art.
Smith: The boundaries about who counts as an artist. You certainly don’t need to go to art school to count as an artist, but it used to be you need some kind of technical skill, you need some kind of physical ability to draw or to sculpt, and technology changes who counts as an artist, and who gets to make art. That happened with the camera, and AI is changing the amount of people in the world who think, “I can do art,” and that’s a very exciting prospect.
Hernandez: All right, now tell me one thing that scares you about AI and art.
Smith: Well, the great middle class of the art world is either artists working freelance for small contracts, or designers working in teams of five or ten, and that’s going to disappear. Freelance artists who can get paid ten dollars to make a little doodle aren’t in as high demand anymore because of AI. And a team of 10 designers working on a project together, a team of 2 or 3 people can do the same amount of work if they’re powered by AI. So, we’re gonna see, I fear, a sort of vanishing of a middle class of artists.
Hernandez: What’s your concern moving forward, you know, when it comes to how students will use the technology? But also, again, because we’ve also seen that, in some cases, copyrighted images are being used, or people are claiming that their images and work is being used, because AI doesn’t know the difference.
Smith: One of the big conversations both on campus today and in particular in this two-day conference we had is something called model collapse. Model collapse takes as its central concern this notion that we’re filling up our academic journals, our research papers, and the internet in general with all this stuff written by bots. And even though GPT can write really convincing scientific papers and code and anything you might care to see, when you fill the internet up with all this GPT-generated stuff, you can’t continue over the next couple years and decades to train new AI on these systems because if you look at the whole internet it’s just a bunch of stuff AI wrote anyway. So you’re not getting better data as you keep training on the internet and that is both a practical concern and an ethical constraint, right? If it’s stealing from copyrighted materials and filling up the internet with things we’re not sure if they’re half copyrighted or how they’re protected, the short answer is, the law doesn’t have a clear decision right now, or it doesn’t seem like it will any time soon, on how you can copyright these AI outputs. And so for my artists, they just want to critique the copyright system, comment on what a silly joke copyright is in the first place. Artists like to sort of poke fun at these questions of appropriation and commodification of art. So I think as the internet gets weirder and weirder and more and more full of AI-written stuff – and by the internet, I also mean these research papers and journal articles – we’re going to have a real crisis of privacy and authorship as well. And it’s more than just a legal question. It’s an aesthetic and an artistic question of, well, what do we actually care about? And that’s where artists come in.
Hernandez: Fascinating. And a little bit frightening at the same time. Griffin, I really appreciate all the information and insight.
Smith: Well, thanks so much for having me. It was a pleasure.
The post As AI evolves, what role will the technology play in the world of art and design? appeared first on TPR: The Public's Radio.