How Making Generative Art Changed My Understanding of Self

Making a generative art collection—the “Medusa Collection,” a set of 2,500 NFTs reframing the Medusa myth—changed my relationship to art. My relationship to art had never, until recently, been a technological one. As an art school student, I made charcoal drawings of still lifes and nudes. As an art dealer in my twenties, I sold paintings, often to collectors who wanted to see the work in person before buying. When I returned to artmaking in my early thirties, I made works on canvas, using colored metal leaf and whole bottles of neon acrylic. My clothes, hands, and hair got covered with fluorescent splatters. I existed in the realm of the analog, the body—and I wasn’t unhappy there.

I fell into NFTs in 2021, quite by accident. However, casual exposure quickly led to deep fascination. Not so much with the technology, but with the culture surrounding it—although the two cannot really be separated. For just as the words we say shape our thoughts, the technology we use molds our relationship with the larger world. After six months of exploration, I knew I wanted to do something ambitious in the NFT space. But it wasn’t until encountering PFP projects like World of Women and works like those by Refik Anadol that I knew I wanted to create something generative.

Generative art, or algorithmic art, dates all the way back—according to Afrofuturist and generative artist Nettice R. Gaskins—to prehistoric basket weaving. Generative art is anything that couples a code, or set of instructions, with a series of artificial events that can output endless variations, be it baskets, paintings, or NFTs. Ellsworth Kelly created paintings that used chance operations to assign colors in a grid. Sol LeWitt and Channa Horowitz made drawings and installations of a similar nature. Before using computers, Vera Molnar—a pioneer of generative art and one of the first women to use computers in her practice—created algorithmic art with nothing more than a pencil and a piece of paper.

However, generative art before computers is a bit like astronomy before the telescope. Certain fields do not come into their own until a tool that bridges the gap between circumscribed human experience and uncircumscribed human imagination comes along. And that, with the advancement of computers and AI, is very much the case with generative art.

Generative artworks—like those by Anadol and any of the artists in the “Artists Who Code” benefit auction and exhibition I’ve co-curated for Artsy, including IX Shells, Ellie Pritts, Cibelle Cavalli Bastos, Morehshin Allahyari, and Aya—are, in my opinion, some of the most relevant artworks being made today.

“I use data as a pigment and paint with a thinking brush that is assisted by artificial intelligence,” Anadol has said of his practice. In his TED Talk, Anadol talks about watching Blade Runner (1982) as a kid and being struck by the scene in which the android Rachel realizes that her memories are not her own but someone else’s—a scene that, in many ways, is the foundation of his practice as he probes “what it means to be an AI in the 21st century.”

Anadol’s work—which applies machine learning to massive amounts of data (up to 77 terabytes) to create mesmerizing visuals that get projected onto architecture—is stunning. Experiencing it indeed feels like watching “machines…dream, remember, and hallucinate,” a stated goal of his. However, I’d argue that the hallucinations he creates are not a machine’s, but his own. The question he is really asking is not what it means to be an AI in the 21st century, but what it means to be a human. Although, the two questions are perhaps not separate—especially as it becomes increasingly unclear where our actual brains and bodies end, and our artificial ones take over.

This takes me back to my own generative art. The “Medusa Collection” started as a series of paintings I made in 2019 after learning that Medusa had been raped by the god Poseidon and turned into a monster by his wife, the god Athena. I made the paintings to work through some of my own traumas and feelings of monstrousness, and, ideally, to connect with others over them. However, the limitations of displaying physical paintings in a physical gallery did not align with the scale of my feelings, nor my need to relate with others.

With the “Medusa Collection,” in which I used an algorithm to create 2,500 unique Medusa NFTs, I was able to blow past these analog barriers. For it, I created six massive Photoshop files, one for each head type. Each file had almost 70 layers, which were separated into seven different categories (background, skin color, shadow color, etc.). An algorithm would then randomly select one layer from each category (except for the “paint gesture” category, in which there could be up to three simultaneous layers) to create a complete, unique Medusa. The 2,500 pieces in the collection are a relatively small sample of the 5.8 billion possible outputs.

The result was a body of work more wild—in terms of color, composition, and scale—than anything I’d made before. By surrendering the creative reins to an algorithm, I was also able to let go of parts of myself that required perfection and planning. After all, there was no way I could possibly troubleshoot all the permutations my algorithm could produce. I had to trust my intuition—along with a force, in the form of an algorithm, outside myself.