Dainotti

Description

Synopsis In 2025, a Google infrastructure deal transforms a Belgian coastal region into a cooling system for AI computation. The flooding is deliberate, a cost optimization where human submersion becomes cheaper than server relocation. A new class of AI-adjacent workers adapts to life beneath the waterline, their bodies modified, their labour invisible. Between prediction errors and system glitches, a blind spot opens. Something uncalculated appears. But is it real, or has the narrator finally learned to see what the models cannot?

Process

I was following the question of how prediction technologies quietly reshape social reality. The urge came from observing existing infrastructure, data centres consuming river water, forecast models guiding policy decisions, and coastal flooding projections, which shift the boundary between describing the future and producing it. I pushed it one step further by asking what happens when forecasts stop being advisory and become operational commands. When a model predicts inevitable flooding, the response becomes accelerated in the name of efficiency. The work comes from a concern about optimization logic treating human bodies as variables inside technical systems. I focus on the moment workers are managed as human infrastructure in support of machine infrastructure. The change is procedural, not spectacular. Corridors are redesigned, a wristband decides who passes a barrier, and a technician reads their body temperature as a metric in the cooling system. Chiptide asks how predictive models forecast reality and then redraw the boundary between living bodies and the systems they serve.

Tools

Chiptide is about AI and was made with AI. I used different systems as production partners, treating their outputs as proposals I could accept, reshape, or reject, letting prediction shape production choices. For images, I built prompt-based workflows in ComfyUI (with different models), Midjourney, Veo, Kling and Firefly (Magnific for upscaling). I facilitated the generation of variations and analyzed which ones the model pushed as the next step. Each shot is both a picture of that world and a trace of the system that produced it. I had no agency in the entire pipeline. Sound followed the same approach Gemini Pro analyzed the sounds after suggesting which sound to look for and the editing, FX, mixing and mastering style. For writing, again, I acted as a sparring partner to six different LLM's. Eleven Lab generated the voices. Across image, sound, and writing, the method was consistent. This mirrors the work's world, where models describe reality and also steer what happens next. Letting tools influence formal decisions tracks where prediction turns into direction, and where judgment and interruption still matter.