Refik Anadol. Sample data visualization of Unsupervised — Machine Hallucinations — MoMA. 2022. Data sculpture: custom software, generative algorithm with artificial intelligence (AI), real-time digital animation on LED screen, sound, dimensions variable. © Refik Anadol Studio

Unsupervised, Refik Anadol’s current installation at The Museum of Modern Art, ponders what a machine might dream about after seeing more than 200 years of art in MoMA’s collection. Anadol uses artificial intelligence to interpret and transform the Museum’s data set—to create new forms that could exist in the archive but don’t, to think about all the paths not taken in the history of modern art and to speculate about what they might be. In turn, the work transforms the environment of the Museum’s Gund Lobby into a large-scale, “living” animation teeming with unique visualizations that unfold unpredictably and have no predetermined outcome. The project involves multiple stages of sophisticated machine learning developed over the past eight years: Anadol’s studio used open-source software1 to search, sort, and classify the publicly available data set of MoMA’s collection, creating a complex spatial map of the archive in 1024 dimensions. The studio then used a machine learning model known as a generative adversarial network (GAN)2 to navigate the map of the archive and, after months of learning, perpetually create new forms with the help of custom rendering software and a supercomputer3. The constantly changing visuals are rendered nearly instantaneously, in shockingly high resolution— the machine “dreaming” of modern art. Visitors are also invited to register for a free digital memento of their visit. We asked Anadol about his art history heroes, what he considers the materials of his art, and what he hopes a viewer will take away from Unsupervised.

Installation view of Refik Anadol: Unsupervised

Installation view of Refik Anadol: Unsupervised

James Turrell. Frontal Passage. 1994

James Turrell. Frontal Passage. 1994

What artists inspire you? Do you see yourself working in any particular lineage of art making?

My work as a generative artist draws inspiration from the legacies of abstraction, systems art, Surrealism, and Expressionism. The works and the incredible visions of the early pioneers of computer art—such as the geometric abstraction of Vera Molnár and the algorithmic drawings of Georg Nees—motivated me to define my own place at the contemporary intersection of art, science, and technology. I am also indebted to the Light and Space movement that emerged in Southern California in the 1960s. Play with optical illusions, Minimalism, and geometric abstraction were its defining features, and I dwell on these elements and strategies frequently in my works. Of course, the fact that the movement was introduced to the public at the famous University of California Los Angeles exhibition in 1971 [Transparency, Reflection, Light, Space: Four Artists] inspired me a lot, as I completed my second master’s degree at UCLA under the mentorship of Casey Reas, Christian Moeller, and Jennifer Steinkamp, and have been teaching there for eight years. With the inspiration that I get from Gene Youngblood’s foundational book Expanded Cinema, and artists such as Helen Pashgian, Fred Eversley, James Turrell, Robert Irwin, Bruce Nauman, Larry Bell, and Dan Flavin, I try to understand and explain the relationship between data, machine intelligence, and space by using cutting-edge light and projection technologies in environments and installations.

What was the first digital art that you encountered? How did it make you feel?

It was The Legible City by Jeffrey Shaw, a groundbreaking interactive art piece [at ZKM Center for Art and Media in Karlsuhe, Germany], where the visitor rides a stationary bicycle through a city simulation surrounded by three-dimensional letters. It brought together many elements of immersive, interactive, and multisensory art that I had been pondering as an undergraduate student at the time and opened new windows for me to think about the city, urban architectures, and collective memories in a different light. When I viewed Shaw’s piece in Karlsruhe, I was already studying with Peter Weibel at ZKM, and I was interested in exploring Pure Data, and later VVVV, both open-source visual programming software platforms. I helped put together the very first media arts exhibition in Turkey, installing 300 artworks from all over the world in 2009. That was when I realized that I wanted to be in conversation with those artists and push the envelope by bringing new perspectives to the field.

What do you consider to be the materials of your art?

Collective memories, archives, light, architecture, generative AI algorithms, software, hardware, code, sound, and, more recently, scent.

Refik Anadol’s studio

Refik Anadol’s studio

“We imagine a future where a symbiotic relationship with machines will give us new insights, knowledge, and the power to not only challenge but change existing systems.”

Refik Anadol

What does your studio look like? Can you tell us about the team that you work with to realize a piece like Unsupervised?

I knew that I wanted to establish a studio when I was an MFA student at UCLA in 2014, and I began doing research about interdisciplinary studio cultures. We started off as a small group with a focus on public art production, but our studio practice expanded into a cross-disciplinary research unit over the years. The Studio is based in LA and comprises designers, architects, data scientists, composers, and researchers from diverse professional and personal backgrounds. We originate from 10 different countries and are collectively fluent in 15 languages (and spill into other cities beyond LA, including Berlin and Istanbul). Our shared dream is to make art for all ages and cultures. The fact that we are a diverse team contributes tremendously to this project. We have been conducting bleeding-edge research and collaborating with leading neuroscientists, philosophers, biologists, medical doctors, environmental scientists, and computational designers in various parts of the world. We use the most innovative methods and most advanced research available to us to challenge received notions and inherent biases in technology, and to imagine a future where a symbiotic relationship with machines will give us new insights, knowledge, and the power to not only challenge but change existing systems, to create a better world.

Installation view of Refik Anadol: Unsupervised

Installation view of Refik Anadol: Unsupervised

Is there anything particular that you’re hoping a viewer will see or sense in Unsupervised?

I think that Unsupervised not only pulls the viewer into a strange world of collective art histories as imagined by a dreaming machine, but also provides a moment of meditation on new modes of perception and sensation. As it unfolds, you can see it speculating about, for instance: How to create an abstract picture. How to render volume and depth in new ways. How to deal with inventing new colors. And even the question, Why?—because these are the problems that artists confronted in the past two centuries.

For this work we used the most advanced generative AI algorithms in the world and created a dynamic, living artwork, meaning that it never repeats itself. At times, it also shows another layer of diagramming its own decision-making paths and correlations. It is based on ethical data research and analysis and has the potential to generate new discourses about how our faculties of perception are changing now that machines are inseparable witnesses of our activities and environments. In fact, we are currently designing a research protocol about the immediate effect of Unsupervised on the viewer by collaborating with neuroscientist Dr. Adam Gazzaley to measure brain signals, heartbeat, body temperature, and skin conductivity at the moment of experiencing the work.

Is there a data set out there you hope to work with someday?

I have been dreaming about compiling the largest rainforest biome data set in the world in collaboration with the Yawanawa people, which would give rise to a real-time, immersive rainforest-inspired artificial reality.

What gets you most excited in art and technology today?

The fact that generative AI algorithms are open to the public is very exciting, and I am watching all the recent developments very closely. I’m enthusiastic about creating new dialogues about how technology is being used outside of its imposed realms and the future of productive transgressions.

Refik Anadol: Unsupervised, organized by Michelle Kuo, The Marlene Hess Curator of Painting and Sculpture, and Paola Antonelli, Senior Curator, Department of Architecture and Design and Director, Research and Development, with Lydia Mullin, Curatorial Assistant, Department of Painting and Sculpture, is on view through March 2023.

  1. UMAP (Uniform Manifold Approximation and Projection for Dimension Reduction) algorithm.

  2. Specifically, customized version of NVIDIA StyleGAN 2 ADA with RAS Latent Space Browser.

  3. Anadol’s studio uses the open-source software VVVV; Pytorch; and the NVIDIA DGX System.