Application Status and Evolutionary Logic of Artificial Intelligence in Smart City Construction and Social Governance

Abstract

Artificial intelligence has transitioned from theoretical promise to operational infrastructure in smart city construction and social governance. This review synthesizes the current application status and underlying evolutionary logic of AI systems across urban domains---including transportation, energy management, public safety, and civic participation---by tracing how technical capabilities, institutional mandates, and socio-technical feedback loops co-shape deployment pathways. Rather than treating AI as a monolithic tool, the analysis distinguishes between reactive automation, adaptive coordination, and anticipatory governance modes, each anchored in distinct data architectures, decision-scope boundaries, and accountability frameworks. Historical progression reveals a shift from isolated pilot projects emphasizing efficiency gains toward integrated platforms prioritizing systemic resilience and equity responsiveness. Key tensions emerge at the intersection of real-time data assimilation and procedural legitimacy, algorithmic scalability and contextual granularity, and predictive modeling and democratic contestability. The evolutionary trajectory is not linear but dialectical: advances in model interpretability spur regulatory refinements, which in turn reshape training-data curation practices; similarly, citizen-led data cooperatives recalibrate municipal AI procurement priorities. Future viability hinges less on computational sophistication and more on the institutionalization of reflexive governance---where AI systems are designed not only to optimize outcomes but to expose, interrogate, and iteratively reconfigure their own normative assumptions. This paper provides a structural map of these dynamics, clarifying how technical evolution is inseparable from political economy, administrative culture, and civic epistemology.