Is Art Dead? And Did AI Kill It?
MoMA's AI video art star Refik Anadol thinks AI will only kill human art if artists don’t adapt
Long before most people got their hands on any kind of AI image generation tool, fine artist Refik Anadol was toiling away in relative obscurity using digital data to craft experimental art that would not be possible without state-of-the-art computational tools. In 2021, Anadol unveiled “Unsupervised—Machine Hallucinations,” an AI video installation at the Museum of Modern Art in New York that displays a constantly moving AI-driven piece trained on the museum’s historic collection of artwork.
This was before AI image generation tools like DALL-E, Midjourney, or Stable Diffusion made it to the general public’s hands. Since then, Anadol has been hailed as one of the unique voices from the world of traditional art paving the way for other artists to begin thinking about how to utilize AI in their own work. After earning a number of awards for “Unsupervised,” MoMA recently decided to add Anadol’s work to its permanent collection.
The debate over what is and isn’t art in the wake of AI’s rise has become intense, particularly in Hollywood. In traditional art circles, from fine artists to comic book illustrators, the debate is similarly intense but less talked about. Many artists who are skeptical (a polite description) of AI image and video generators take pride in the callouses on their hands and the many sleepless nights they’ve spent over a lifetime developing skills that AI can now, at least in some cases, mirror with powerful effect. I spoke with Anadol about his work, and how he thinks about AI in a way that might help other artists struggling with this seismic shift in visual expression.
Adario Strange:
Unsupervised…why did you give your project that particular name?
Refik Anadol:
Unsupervised is the curatorial name devised by our wonderful curators, Michelle Kuo (MoMA’s Marlene Hess Curator of Painting and Sculpture) and Paola Antonelli (MoMA’s Senior Curator of Architecture and Design and Director of Research and Development), but also with me as well. [We used it] because unsupervised is a technique for AI research, as humans are less included in the process of machines deciding things. For example, in our project, we let the machine look at the entire MoMA archive with a much more unbiased point of view. I guess the name represents that we were in uncharted [waters].
Strange:
Before this project, what was your primary medium of art? Was it painting, or sculpture, or has most of your work been data-driven computer art?
Anadol:
I don't think I'm talented in drawing, but I know how to draw in my mind, mathematically. I love computation. And I’ve loved computers since my childhood. In 2008, I think I coined the term “data painting.” I started using custom software to create artistic expressions. I always use light as my primary material. This can be projection or LED technologies, basically any light source.
Additionally, seven years ago, I was the first artist-in-residence at Google, where I started my AI journey in a group called Artists and Machine Intelligence. It's an incredible program. Over the years, our entire focus has been on AI, data paintings, data sculptures, and immersive environments.
The mind-to-machine transformation is super inspiring, and productive as well. For example, the data we’ve been using over the years. We use brain activities, heartbeat data, orbital space data, weather conditions, and Wi-Fi Bluetooth signals, it’s a very diverse palette of data to use. And each data set offers a unique approach. So in this context (“Unsupervised”), we used the beautiful artworks in the MoMA collection as a different approach.
Strange:
Although the whole AI art generation journey kind of started around 2014-2015 and has gradually developed to where it is now, as recently as 2021, most people didn't have access to these AI art generation tools. Now that so many people have access via tools like OpenAI’s DALL-E, Midjourney, and Stable Diffusion, what do you think you've learned by having a bit of a head start on many other traditional artists?
Anadol:
When I started training neural networks using custom data sets, to me, that was the inspiration. I was not inspired by the idea of “let's put everything here, and let's make an everything mission.” Instead, I thought, we have to be curating data for specific reasons.
For example, if I talk about nature, let's download data on all the national parks in the world, or all the trees, or all the flowers in the world. I love being specific and I love specific domain data research. But over the years, of course, DALL-E and clip guidance diffusion models have become super interesting. But again, if you [input] the same sentence, you have the same output. I don't believe [the user is] bringing anything artistic, they're just giving an idea and inspiration for the creative. I love prompt engineering, but I don't believe the output is art. The art I think starts right after the outputs. What do we do with that? How do we transform these outputs into a creation?
But at the same time, I'm super grateful for these [new AI generative] tools. I mean, it's beautiful. It's amazing. But anyone with the same prompt can get the same thing. I feel like there's much more potential once you have your own AI model and you craft it.
Strange:
What if a person isn’t technically savvy enough to train their own model and they just use the off-the-shelf tools? Right now, people are using mainstream AI generation tools and tagging their artwork as AI according to the software they used (#midjourney, #stablediffusion, etc.), but I suspect soon they will just stop tagging their work as AI, and this could blur the lines between people who created art by hand and people who are just outputting AI generations. As a fine artist, what do you think about that prospect?
Anadol:
I think it’s liberating. I love how human creativity can be shared as freely as possible. As an artist, I love when the physical and the virtual blend. I think it’s a beautiful problem. [Speaking] from my inner voice, I love that we are not able to like understand what is going on. And eventually, in a world where AI knows everything, and it can dream, who will define what is real? This was my question several years ago during my residency at Google. But at the same time, I'm still coming back to the same position: I believe that it's just becoming another search box. The search bar for creativity.
Strange:
Is art dead? If anyone can imitate any artist just by typing in a few words, and they don't label their art as AI…what is the future of art? What happens next?
Anadol:
The first feeling I think many, many creators are getting is the feeling of an extension of the mind. When I use these tools, I feel like my mind is expanded and enhanced, giving me something I knew was there, but I couldn't bring up. From a cognitive context, my mind is extended and enhanced by using the tools. My guess is that it will be just another mental tool, in the long run, that allows you to just enhance your imagination.
Strange:
Earlier you indicated that when someone types a text prompt into one of these AI models, the output itself is not art.
Anadol:
Not art, creation. These [AI-generated] creations are not art, yet. To me, art is a process on top of when these outputs arrive in the hands of the creative. There is one more process, I think, that is needed to be able to contextualize the output of the machine.
Strange:
So you're making a distinction between creation and art?
Anadol:
Correct. Art is like philosophical discourse, context, and imagination, and that’s not always up to the universe.
Strange:
Okay, so just to be clear, in your view, creation is the craft of bringing something to be, whereas art indicates intention and meaning.
Anadol:
Correct. Having AI as an extension of the mind is an incredible process, and it can unleash the creativity of creators. But for artists, to turn it into a true value of an experience in life, through the use of wisdom or experience, or an artifact in a journey, I think that's where another layer of imagination, another layer of creation, becomes part of the journey.