In Nora Al-Badri's Babylonian Vision, a pretrained neural network generates the so-called »techno-heritage.« This speculative approach to archaeology is based on the algorithmic recombination of images of Mesopotamian, Neo-Sumerian and Assyrian artefacts. These have been collected through web crawling and scraping the digital collections of five major museums. What emerges are AI-driven object visions that act as the subconscious of a collective memory. The violent colonial patina surrounding much of the cultural loss in the regions of today’s Iraq is thereby countered with a fictive, generative, and future-oriented object production.