Latent Heat Generation is an artistic research project investigating the hidden material infrastructures of AI systems through thermal imaging feedback loops, creating real-time feedback where computational waste heat becomes generative material for AI imagery. Developed from the vantage point of a Taiwanese artist, the work emerges from a personal dilemma: being limited by computational power while simultaneously being connected to the very infrastructure that enables AI development globally.
The installation reveals multiple forms of generation: thermal energy production, AI image creation, semiconductor manufacturing generations, and generational politics of technological dependency. Custom software converts thermal readings into prompts for stable diffusion models, creating recursive loops where GPUs examine their own material conditions through heat visualization.
The thermal investigation reveals how computational heat carries signatures of global manufacturing dependencies, questioning the apparent immateriality of digital processes by making visible the energy transformations and geopolitical relationships embedded within AI systems.
