From performative to predictive-performative design: A review of current trends in performance-based design and their impact on urbanism.

This article focuses on the emergence and development of Predictive-Performative Urban Design (PPUD), a new paradigm in performance-based urban design enabled by advances in artificial intelligence (AI) techniques such as machine learning (ML) and deep learning (DL). PPUD integrates predictive models with generative design and real-time simulation tools to optimize urban design alternatives rapidly, addressing limitations of earlier performance-based approaches that relied heavily on deductive mathematical models. The paper traces the historical evolution of performance-based urban design from deductive and parametric methods to the current predictive-performative approach, highlighting key research projects at the Massachusetts Institute of Technology (MIT) and the Austrian Institute of Technology (AIT) that exemplify this shift. It also identifies ongoing challenges related to improving prediction accuracy, expanding the scope of predictive tools, and validating models across diverse geographic and climatic contexts, emphasizing the potential of PPUD to enhance urban resilience and climate adaptation through more responsive design practices.