Most of the language around AI and creativity right now treats the machine as a tool — something a human directs, prompts, or corrects. Sougwen Chung, a Chinese-Canadian artist and researcher, has spent close to a decade arguing for a different word: collaboration, chosen specifically because it implies mutual exchange and what she calls “relational risk” — something neither party fully controls.
Watch: Sougwen Chung — Why I Draw With Robots (TED) (YouTube)
The work behind the argument
Chung’s primary body of work, the D.O.U.G. (Drawing Operations Unit: Generation) series, spans multiple generations of a bespoke drawing system she works alongside rather than through. The series has been presented at the Victoria and Albert Museum — where a piece entered the museum’s permanent collection — the National Art Center in Tokyo, and the World Economic Forum in Davos. Her newest piece, Recursion 3, is a 10-meter scroll built from her own brainwave data, being completed live at Art Basel Hong Kong’s Zero 10 sector this year.
Why “I’ve had to become machine-readable” is the actual thesis
Chung has described the core tension of her practice bluntly: to collaborate with these systems, she’s had to make her own creative process legible to a machine — translating gesture, intention, and style into something a system can respond to. She frames this as an existential tension core to contemporary life generally, not just her own studio practice: the machine doesn’t just extend what she can do, it reflects her own choices and biases back at her, which is a very different relationship than treating a generative model as a faster paintbrush.
Why this framing matters right now
As generative AI tools get folded into nearly every creative discipline this year — video editing, music production, 3D modeling, shader authoring — most of the marketing language around them defaults to “assistant” or “co-pilot,” framing that quietly implies the human stays fully in control and the machine simply executes. Chung’s decade-long insistence on “collaboration” and “relational risk” is a useful corrective to have in mind: her work argues that the more interesting and more honest version of human-machine creative work involves the machine genuinely changing what gets made, not just accelerating a plan the human already had.