Inverting the language learning model hallucinatory methodology, we implement an interactive dialogue to coax out the machine's knowledge gaps, directing the temporal resistances in and around the filmed photographs. We intervene in 'ground truths' with our custom (human-written) inferences developed as text, soundscapes and resistant images, which interrogate the hallucinatory inferences of large language models. Our interventions symbolise resisting automation and disrupting renderings of spatial and temporal patterns that sustain the current planetary condition.
The development method involves several overlapping and back-and-forth processes between analogue and digital, 2D and 3D, code and tools, image and text.
So to ensure coherent pieces and to render our concept clearly, we set some constraints for the process:
- One shot of the photographic frame
- Using immediate generative outputs
- Intuitive non-linear edits. Repetition as a mode of being.
- The wilder variable, a series of animations as human and inhuman resistance