Evolving from Rules to Learning in Urban Modeling and Planning Support Systems

Urban modeling is being reshaped by advances in artificial intelligence (AI) and data-rich sensing. This review assembles an integrated evidence base connecting spatial dynamic modeling (SDM), planning support systems (PSSs), urban analytics, and governance concerns. We analyze 1290 publications (2000–2025) using a reproducible pipeline that combines structured literature retrieval with retrieval-augmented generation (RAG) for semantic screening and evidence extraction. Bibliometric mapping and a rigorous coding framework structure the synthesis. The results reveal three linked trajectories. First, SDM has progressed from rule-based simulation toward learned spatial representations using deep and multimodal learning. Second, PSS has evolved from static analytical tools to interactive and participatory environments that embed AI for scenario exploration and stakeholder engagement. Third, governance themes such as transparency, fairness, and accountability have gained importance but remain unevenly implemented in modeling workflows. Building on these findings, we advance AI-aligned SDM, which integrates explainability, uncertainty reporting, documentation, and participation into model design to strengthen institutional accountability and evidence-based planning. A forward research agenda emphasizes methodological fusion between simulation and learning, institutional design for continuous model stewardship, and epistemic pluralism connecting local knowledge with AI to advance equitable and transparent urban governance.