The Process Behind BMW’s Artificial Intelligence Artwork From Frieze New York 2021

Joining the illustrious ranks of Warhol and Lichtenstein, Jenny HolzerCao Fei, Futura and and dozens of other marquee visual artists, artDrunk founder Gary Yeh and creative technologist Nathan Shipley (of Goodby, Silverstein & Partners) have contributed the “Ultimate AI Masterpiece” to BMW‘s lineage of art collaborators. Unlike their predecessors, Yeh and Shipley’s series of art-adorned BMW 8 Series Gran Coupes debuted virtually. And, the two collaborated digitally from Seoul and San Francisco, respectively.

To dress the cars, Yeh and Shipley fed 50,000 images of classic artwork, created across some 900 years of history, into NVIDIA’s open-source software StyleGAN. They coupled this with a curation of 50 works from international contemporary artists who have partnered with BMW. With all of this visual information, the artificial intelligence generated several new images, which Shipley and Yeh mapped onto the vehicles. The mesmerizing final results (which were revealed through an official online viewing room at Frieze New York, in alignment with the 50th anniversary of BMW Group Cultural Engagement) attest to the artistry within technology. While each abstract artwork has been informed by the history of art, they are undeniably contemporary.

“In this uniquely isolated time, we wanted to play on this idea of cars and mobility to see how we can continue to connect people around the world despite closed borders,” Yeh tells us. This was what led them to include the 50 works by international artists. “Virtually all of these contemporary artists bring their local culture into their work or deal with global topics like North and South Korean craft, the African diaspora, and the environmental impacts of plastic bags. The range of their aesthetics then felt like teaching the AI new techniques, adding to its existing toolbox of 50,000 other art history images. With living artists always looking back on art history for inspiration, the AI works similarly in pulling from across these thousands of images.”

Regarding the database at the foundation of it all, Shipley adds, “The art history images encompass a very broad range of artwork. The thinking here was to cast a broad net and was part of an exploration to see what happens when we ask the AI to learn, generally, what is art? While we explored smaller datasets, we ultimately intentionally didn’t exclude things as our starting point. It is very striking to watch the generated imagery morph from one style and genre to another.”

“The model created by the AI about the history of art then became its starting point for when we showed it the 50 contemporary works,” he says. Shipley and Yeh were able to observe as the AI began to “understand” art, and transform “from this broad range of historic styles to something more specific.” Shipley provided a GIF (below) to showcase this progression, from “the history of art to Lee Bae’s work over the course of training.”

“I was amazed by how the artwork constantly evolved, as if you were seeing the AI learn and create in real time,” Yeh adds. “I could pick out certain elements from the artists that were selected, but was then inspired by how such new forms and color combinations were created.” Moreover, he was impressed with its ability to map over the vehicle. “I was especially happy to see how the AI-generated work wrapped around a car, emphasizing that sense of movement through space and cultures that we were after in selecting these artists.”

This virtual installation pushes the boundaries of what BMW has done in the cultural sector for the last 50 years—and aligns with recent fervor over digital art.

Yeh says, “Taking on the legacy of the BMW art car, the ‘Ultimate AI Masterpiece’ adapts to this new environment, playing to new platforms like Frieze OVR. With past projects that have involved video art and AR, working with AI at BMW is also a natural evolution of finding and supporting artists that are staying current with developing technologies.”

These vehicles also beg the question: can AI once day be creative on its own, or will the hands of the creators always influence the artistry? “At the moment,” Yeh says, “AI is used more as a tool that artists, engineers and curators can use to push the boundaries of their own work and research. That’s the beauty of this collaboration. We come from different backgrounds in art and tech, but ultimately found the intersection where the most exciting aspects of each discipline could be highlighted.” Shipley adds that there were thousands of human choices along the way that led to their particular results, which include setting the parameters of the neural network that did the learning.

“This tension between technology and traditional craft is something I often wrestle with and try to find a balance between,” Shipley says. “The idea of generative artists working directly with classically trained artists is fascinating to me though. I would love to see more painters, illustrators and sculptors working with these tools. That feels like the right balance to me.”

Creativity intertwines with consciousness—and although a neural network generated the imaginative art here, it’s the collaborative vision of Yeh and Shipley that brings it to life in an extraordinary way.

Images courtesy of BMW

Read More

David Graver