The Urban Intelligence Model from AI-Enabled Insight to Public Action and Institutional Learning

Abstract

Cities use artificial intelligence, urban data platforms, and Digital Twins to perceive conditions and anticipate risks. Yet analytical capacity does not by itself ensure legitimate decisions, coordinated intervention, evaluation, or institutional learning. This article asks how the literature represents the transformation of urban data into public action and to what extent existing frameworks integrate the capabilities involved. A scoping review retrieved 2,209 records from Scopus, Web of Science, and IEEE Xplore, producing 1,397 unique records after deduplication. Title and abstract screening mapped the field, while a full-text comparison examined the frameworks that directly challenged the proposed model. The review shows that individual capabilities and recursive lifecycles are increasingly integrated, with the closest challengers approaching an end-to-end architecture. The remaining problem is less the absence of stages than the continuity between technological, organizational, and institutional functions. In response, the Urban Intelligence Model differentiates seven recursive capabilities - Perceive, Interpret, Anticipate, Deliberate, Act, Evaluate, and Learn - and identifies three institutional boundaries where intelligence fails to become legitimate action or cumulative learning. An illustrative application to urban flooding in Recife demonstrates the model's diagnostic use. The contribution is a domain-independent conceptual architecture whose empirical validity and operational maturity require subsequent testing.