Sea Oasis – Survival Architecture investigates how generative AI can helps scientists preserve and encourage biodiversity by predicting the performance of restoration interventions. It tackles Mediterranean marine species depletion by submerging algorithmically designed 3D-printed oyster reefs, whose outcome in proliferating life are monitored by the University of Padua. By managing natural and anthropogenic variations, predictive generative AI suggests ecosystem simulations for efficiently planning reef restoration interventions and facilitates the evaluation of their impact. The installation illustrates this process through different scenarios, dynamic data visualisation, and by showcasing 3D-printed modules.