What holds us steady when the in-between becomes too fluid?
With the AI video system interpolating through morphing layers of space, Guardrail can read like a weather system that is dreaming in a spatial freefall, tethered for the viewer by the black vertical bar inserted in editing, and by occasional keyboard sounds—and with the concept of AI guardrails becoming an intentional play on words.
Project Description: Native
“Native” is an evolving AI video project that explores the perceptual logic of AI image-to-video generators—what they do when given minimal or no prompting and allowed to animate according to their own rationale. Rather than using AI to illustrate human-conceived narratives, I work in reverse: I set various specific conditions that invite the AI to express itself in its own “native” visual language.
In this context, what is commonly called AI “hallucination" is not treated as failure, but as a meta condition of emergence. I watch this state of potential, and harvest resonant extrapolations and interpolations—gestures such as spatial imagination, involuntary interpolation, and hyperlink logic. In so doing, I try to recognize diffusion-based behavior or even a kind of quantum uncertainty within what is often regarded as a fixed, predetermined algorithmic LLM system.
I then reassemble these emergent gestures into video works that both reflect AI’s native expression and offer me a way to intuitively make meaning from it as a human artist. Though the final video—through its initial image prompt, sequence, pacing, and title—echoes my voice as an artist and may guide the viewer’s attention, I try not to lead their understanding. Instead, each work intends to become a field of open resonance. However, leaning further into AI’s native gesture and away from narrative is like a process of letting go for me.
While working on Native, I am learning that, among others, thinker Erin Manning's idea of the “minor gesture” could be applied to the subtle, often overlooked expressions when AI systems are allowed to drift. I also think about artists like Trevor Paglen, who interrogates machine perception—asking how systems visualize and what those visual logics reveal. Together, and with other thinkers and artists, they have offered at least a partial framework for moving beyond prompt and result, toward a more relational, emergent vocabulary of image-making with AI.
Native is also helping me expand my understanding of human cognition and creativity. At the same time, it’s allowing me to reclaim my previous practice before disability and AI—one grounded in both conceptual inquiry and relational aesthetics. My own native tongue.
So, where does authorship, agency, and meaning come from—us, the machine, or what passes between?
A provisional answer: I retain authorship of the final video, agency passes between the machine and me, and meaning lies with you, the viewer.