For decades, I’ve understood my role as an artist as a kind of steersman. Adrift as we are on the waters of technological change, artists are uniquely positioned to see the direction it is going — perhaps, straight into a pile of craggy rocks — and to drop an oar into those waters in ways that shift the course.
The technological imagination comes to us through millions of dollars of advertising and lobbying by the tech firms and manufacturers. The social imagination needs a voice, too, and it is often articulated through projects by the artists navigating and negotiating against that technological current.
There are a number of definitions of “AI art” floating around us. Some see AI art as the driftwood of culture, ready to be picked up and transformed into something useful and perhaps even new. Others see nothing less than the debris left behind by a shipwreck — a pirate’s ship of stolen data, illicitly grabbed away from artists and illustrators. I see no contradiction between those two positions, but maybe because I come from the steering tradition. The steadiest course is somewhere between the two extremes.
My work with technology always has two, mutually reinforcing themes. I want to understand the technology, and I want to say something about it. Often, making things is how I arrive at what I want to say. The founder of video art, Nam June Paik, was famously skeptical of video. It was at once a source of great possibility, but also great power and control. “I work with technology to hate it properly,” Paik said, reflecting on the ambiguity of the artist. At once, we’re observers of the shifts in culture technology brings, and ambassadors of that coming change.
Artificial Intelligence is a peculiar technology for an artist, because there is little misuse or appropriation left open to us. The technologies we discuss so passionately are designed for art making. On the one hand, Large Language Models seek to replace the creative expression of choosing words with the more editorial process of refining and shaping words that it provides. Diffusion models — tools used to generate images, such as DALL-E 2, Midjourney or Stable Diffusion — create images and illustrations wholesale. Capturing art from artistic technologies is a surprisingly daunting task. Do what it says in the instruction manual — type words into a window and produce some images — and you are simply using the product. As an artist oriented to the creative misuse and even abuse of technology, doing what I am told is an extremely dull challenge.
Most technology that artists have appropriated in the past were incidental to art making. To repurpose them was itself a creative gesture. Edison envisioned the wax cylinder for business meeting memos, not for musicians. Paik found TV’s could be distorted with magnets and synthesizer signals. Even Duchamp transformed the way we look at urinals.
Today, artists who want to interrogate AI tools are in a unique position. The companies that produce them have already decided that they are art. To engage means to agree with them. But it does not mean we have to agree with them on their own terms. I am interested in how we can appropriately appropriate art-making tools for making art. How might we make the art we are not intended to make?
A number of my works respond to this concern. Shared here is one of them, a short documanifesto using AI art tools to tell the story of how the tools work. It aims to reveal and complicate many of the overlooked problems inscribed into the logic of these systems, particularly when it comes to their labelling and categorization of the world.
It is a manifesto in that I am also aiming to find ways to break that process apart using the tools themselves. Creative misuse motivates a spectrum of applications, such as confusing the system by asking for images it is technically incapable of producing.
In the end, I understand that many are angry or skeptical about these tools. I agree. Others see them as the opening up of a new form of media art — and I agree. To make sense of the contradiction, it helps to step away from the mythologies and promises of these tools and to look at them for what they do, how they work, and how those core facts can be decontextualized and applied in novel ways.
To learn more about Eryk’s work visit his website at https://www.cyberneticforests.com/