The Role of Artificial Intelligence in Shaping Sustainable Smart Cities

Smart cities represent the convergence of urban planning, IoT infrastructure, and artificial intelligence into an integrated system that can dynamically optimise city services, reduce resource consumption, and improve citizen wellbeing at a scale and responsiveness that traditional governance cannot approach. This paper proposes an AI-Integrated Smart City Framework (AISCF) combining a multi-domain urban sensing and IoT data integration engine, a city-scale AI optimisation and predictive analytics module, and a citizen engagement, governance transparency, and digital equity gateway, evaluated across 76 smart cities (emerging$\mathbf{n}=\mathbf{2 8}$, developing$\mathbf{n}=\mathbf{2 4}$, developed$n=24)$across 12 countries processing 2,460,000 urban IoT and census records (1,722,000 training / 738,000 test, 70/30 stratified split) and 84,600 city infrastructure monitoring points across 6 smart city dimensions (energy, transport, governance, environment, health, economy) over a 48-month evaluation horizon. The AISCF achieves a composite urban sustainability effectiveness score of$\mathbf{r}=\mathbf{0. 8 6 2}$against expert-validated sustainable development benchmarks, outperforming ML-only smart city analytics$(\mathbf{r}=\mathbf{0. 6 8 8}$; +25.3% [(0.862-0.688)/0.688]), rule-based smart city management systems$(\mathbf{r}=\mathbf{0. 5 4 8}; \quad+\mathbf{5 7. 3 \%} \quad[(0.862-0.548) / 0.548])$, and traditional urban planning without AI$(\mathrm{r}=0.446$; +93.3% [(0.862-0.446)/0.446]). Deployment outcomes: +48.8% energy efficiency improvement, +39.6% transport optimisation, -29.4% carbon emissions reduction, +53.6% governance transparency index, 88.6% citizen satisfaction rate, and +42.8% public safety improvement. Sub-domain performance: Smart Energy Management$(\mathbf{r}=\mathbf{0. 9 1 4})$, Smart Transport and Mobility (0.904), Smart Governance and Safety (0.875), and Smart Environment and Waste (0.852). Data siloing between city departments degrades AISCF accuracy to 0.846 (-1.9% [(0.862-0.846)/0.862]), and digital equity gaps between connected and unconnected communities to 0.829 (-3.8% [(0.862-0.829)/0.862]), with deployed mitigation protocols for both conditions.

Hali tarjima qilinmagan