Generative Metaverse is a workshop that revisits the relationship between spoken and written language and architecture as well as current notions of the metaverse. Generated through a combination of the latest machine learning models, images can be generated from text inputs. The parameters behind each model and semantics of these inputs establish new forms of architectural and cultural production where human agency remains critical in this creative process. In this workshop, we will question whether this novel design process would pass the architectural Turing Test, a question posed in last year’s Machine Intelligence Workshop. To do this, students will learn about text-to-image machine learning models, including CLIP, Dalle, and Imagen, and image-to-3D processes in Grasshopper, and apply these to architectural design outputs. We will collectively create a Latent City of the Captive Globe – a series of latent urban morphologies and students will learn an overview of how 3D NFTs can be minted, as a digital and personalized metaverse.