Generative AI in Urban Planning: A Systematic Review and Human–AI Co-Creation Framework for Responsible Urban Futures

Generative artificial intelligence (AI) is increasingly used in urban planning to generate land-use scenarios, mobility models, digital twins, participatory visualisations, and design alternatives. This PRISMA-guided review synthesises 63 peer-reviewed publications and selected scholarly preprints from 2015 to 2025. It maps six model families, five application domains, and five human roles: initiate, constrain, co-create, evaluate, and deploy. Findings show strong emphasis on problem framing and constraint-setting, but limited evidence of iterative co-creation, human-centred evaluation, governance, or deployment. To interpret this imbalance, the article develops role-complete AI urbanism for maintaining human judgement and institutional accountability across the full workflow.