This artwork explores the limits of visibility in an era of image saturation. Using thousands of images sourced from Google with terms such as ‘pollution,’ ‘deforestation,’ ‘wildfires,’ ‘CO₂ emissions,’ ‘floodings,’ and ‘glaciers melting,’ an automated pattern recognition system—a GAN—was trained to generate new visuals. The resulting images fluctuate between abstraction and figuration, questioning whether abstraction can make the overwhelming and often ignored imagery of environmental collapse visible again.

Paradoxically, the very tools used to create this artwork are themselves entangled in the cycle of pollution and CO₂ emissions. The computational power required to train the GAN contributes to the same environmental degradation it seeks to critique. In a cyclical process, the artwork ultimately mirrors the mechanisms of AI overproduction, generating images at an exponential scale until they collapse into excess once more. As the visuals multiply exponentially within the frame, they once again lose their meaning, dissolving into a static, monochrome colour.