Geospatial Artificial Intelligence (GeoAI) for Sustainable Urban Governance: A Systematic Review and Governance-Oriented Framework

Abstract

GeoAI has advanced the analysis and prediction of urban systems by integrating GIS with artificial intelligence, yet its contribution to urban planning and governance remains limited. Many applications function as black-box predictive tools, making results difficult to interpret, justify, audit, and use in formal planning decisions. GeoAI therefore remains a detached analytical tool rather than an integral component of urban governance. This review examines the governance of computational spatial analysis in planning, with GeoAI as the emerging case in which requirements for transparency, accountability, and participation become most acute. A systematic search of Scopus on 5 June 2026 retrieved 806 records, of which 32 studies met the eligibility criteria and were evaluated thematically, complemented by seven illustrative cases. The analysis addresses three dimensions of governance: regulation, ethics, and participation. Across these dimensions, the study identifies common challenges related to explainability, auditability, transparency, accountability, data governance, policy alignment, and stakeholder access. Only 8 of the 32 studies apply an AI or machine-learning technique directly and 11 contain no learning component, while the 7 cases are purposively selected illustrations rather than a systematic comparison, only one of which documents an implemented learning method. The paper proposes the Governance-Oriented GeoAI (GoGeoAI) Framework for Sustainable Urban Governance, designed to support the transition from predictive GeoAI toward governance-oriented decision support through model explainability, auditable decision logic, policy translation, data governance, and participatory interfaces.