Errant Standard, 2025
4k video loop, custom python script, TRT: 2:37, generative captions continuous and variable

By engaging with generative systems in deviant and convoluted ways, I hope to expose their limitations and biases, a divide between meaning and representation and a divide between future tools and users. This process aimed to highlight the detached, vaguely recognizable “vibe” of an object but not an “understanding” of its use. Through back and forth exchanges with a local LLM and Stable Diffusion Imaging, I imagined a “new” tool, one for the hand, and its evolution across epochs of fictional users. 

The resulting handmade, digitally animated representation of this tool appears to showcase layers of histories, labor, use, and re-use. Of course, it is a palimpsest of existing ontologies of existing tools; as we understand it, generative “AI” cannot imagine anything new.

I reintroduced an LLM as an observer of this now articulated, contextualized object and its virtual materiality, texture, and environment. I wanted to see if the system could understand or conceptualize the tool’s use, inherent purpose, or ready-at-handness. The real-time generated captions testify to the AI’s struggle to conceptualize the tool’s practical use, instead subsuming it within a broader narrative framework.

This narrative entanglement of human effort, machine processes, and ideological systems mirrors the creative, but mediated processes of 3D modelling, animations, and writing central to the video. Animated as a seamless, ever changing, but also not changing, narrative loop, “Errant Standard” critiques the capacity of tools conjured from one future of a cyberlibertarian, technoutopia, perpetuating exploitative practices and maintaining fictions of their liberating potential, while demanding a confrontation to the pedagogy with which we surround technology.