Human-AI Co-design Framework for Visualizing Lost Ancient Cities : The Case of The Egyptian Delta Beyond the ruins

This research develops and tests a human–AI co-design framework for the responsible speculative visualisation of Imet (Tell Nebesha), a radically underdocumented ancient city in the Egyptian Delta. Addressing the limits of conventional digital heritage and the bias of generative AI toward monumental, romanticized imagery, the study examines whether non-experts can produce plausible, evidence-grounded visual hypotheses while keeping uncertainty visible. Across three iterative workshops, immersive prototyping, and an external expert evaluation, participants generated speculative outputs for Imet, refined them through archival evidence, and translated them into multiple simultaneous hypotheses in immersive environments. The findings show that non-experts, including those with little prior knowledge, can contribute meaningfully to speculative heritage visualisation, that evidence improves the plausibility of outputs, and that presenting multiple hypotheses communicates uncertainty more effectively than a single reconstruction. The study also shows that immersive media can convey rather than conceal uncertainty when carefully framed. Conceptually, the thesis reframes uncertainty as something to design with rather than merely disclose, and it advances community-realised heritage as a model in which diverse publics, including a source community, collaboratively produce and critique visual interpretations of the past. The research does not claim to reconstruct Imet, but to establish a defensible and shared way of imagining what has been lost.