Cities are complex, dynamic environments, requiring huge numbers of services and systems to facilitate and better the lives of the citizens within them. Keeping up with the demands of modern life has led to the creation of Smart City Digital Twins (SCDT), which are complete and bidirectional Cyber-Physical Systems (CPS) acting as observation and control mechanisms. Current SCDTs are typically bespoke implementations, catering to the city's unique needs and footprint. Generative AI will enable the generation of broader possible visions of the city, but the current data created by SCDTs is insufficient to train generative AI. This is a common problem for AI, and synthetic data is utilised to augment the training set. This paper proposes a novel concept for the creation of synthetic data; the use of the Meta, Character, Spatial Artificial Intelligence (MCS-AI) Model to emulate and therefore build the vast amounts of synthetic data required for a City Generative AI.