How AI is reshaping smart cities across governance, mobility, and more pillars

Artificial intelligence (AI) is reshaping smart cities worldwide, offering evidence-backed insights across six interconnected domains - governance, economy, mobility, environment, living, and people. A new study “Artificial Intelligence for Smart Cities: A Comprehensive Review Across Six Pillars and Global Case Studies” examines over 90 scholarly sources and case studies to present a structured assessment of AI’s real-world applications, technological enablers, and ethical risks in urban settings.

The authors synthesize global practices with bibliometric and thematic analysis to assess how AI technologies, ranging from predictive analytics to edge computing, are improving urban functionality and resilience. Drawing from cities such as Singapore, Amsterdam, Delhi, and Toronto, the study, published in Urban Science, not only documents successes but also highlights policy gaps and challenges, such as digital inequality and fragmented governance.

How is AI being applied across smart city domains?

The study categorises AI applications into six “pillars” that represent the functional cores of modern smart cities: governance, economy, mobility, environment, living, and people. In the governance domain, AI enhances service delivery through e-governance platforms, digital identity systems, and multilingual chatbots. Estonia’s e-Residency program, for example, provides secure digital access to services across borders using blockchain and AI authentication, while Singapore’s "Ask Jamie" chatbot handles multilingual queries for over 70 agencies, boosting public engagement.

In economic infrastructure, AI drives smart building automation, energy optimization, and startup ecosystems. The Barcelona 22@ district, home to over 1,500 AI-driven startups, contributes €3 billion annually to the local GDP. AI-integrated smart buildings have also improved energy efficiency by over 99%, using reinforcement learning and predictive maintenance algorithms trained on real-time sensor data.

Mobility is being revolutionized through connected and autonomous vehicles, edge-based routing systems, and Mobility-as-a-Service (MaaS) platforms. London uses AI to manage congestion through predictive traffic modeling, while Helsinki’s WIM platform enables multimodal transport integration and climate footprint tracking. These systems reduce emissions and promote sustainable transport behaviors.

The environmental pillar showcases AI’s role in real-time monitoring of air and water quality, energy management, and climate resilience. Amsterdam’s smart grid reduces peak demand by 25%, and Delhi’s SAFAR system predicts air quality 72 hours in advance using machine learning. Advanced models like WT–CNN–GRU hybrids also demonstrate high-accuracy water quality forecasting.

In the smart living domain, AI and IoT converge to enable autonomous home environments, predictive healthcare, and embedded learning systems. Toronto’s Smart Health Clinics reduced emergency visits by 18% using AI diagnostics, while Dubai’s smart homes provide personalized AI learning platforms under the Smart Dubai 2030 agenda. Secure AIoT protocols such as SAIoT-SL safeguard these systems with low computational overhead and formal verification.

The “people” domain focuses on empowering citizens through personalized learning, AI-driven civic platforms, and public sentiment analysis. Singapore’s SkillsFuture platform offers adaptive upskilling based on AI recommendations aligned with future labor market demands. Meanwhile, edge AI and 6G digital twins are enhancing cybersecurity and real-time service responsiveness in civic applications.

What are the barriers to AI  implementation in urban contexts?

Despite these benefits, the study identifies three interlocking barriers: data privacy, ethical governance, and infrastructure inequality.

First, data privacy remains a foundational concern, particularly in centralized architectures vulnerable to breaches or misuse. While Europe enforces strong protections through the GDPR, other regions like China have adopted aggressive surveillance models without adequate safeguards. Federated learning and blockchain integration are proposed to balance innovation with decentralized data security.

Second, ethical risks are surfacing around algorithmic bias, automation-led job displacement, and opaque decision-making. Instances in the U.S. and U.K. reveal that AI systems used for policing or welfare distribution can exacerbate systemic discrimination. Amsterdam offers a promising countermodel with its citizen-led AI ethics panels and transparent review boards.

Third, infrastructure disparities are hampering implementation in many mid- and low-income cities. Nairobi and Dhaka, for example, suffer from limited broadband and outdated systems, making large-scale AI deployment impractical. Even developed cities like Barcelona face difficulties expanding AI tools to peripheral neighborhoods due to legacy transport networks. The authors recommend strategic investments, public–private partnerships, and locally adapted funding models to close these gaps.

The study also documents failed or controversial initiatives. Toronto’s Sidewalk Labs project was terminated over concerns about data governance and surveillance. San Diego suspended its AI-powered streetlights after backlash over consent. In the Netherlands, an AI system used to detect welfare fraud was legally dismantled for targeting marginalized groups unfairly. These examples underscore the importance of human-centered design and rigorous regulatory oversight.

What strategic actions can cities take to build resilient AI-driven futures?

The paper outlines several strategic recommendations to ensure AI enhances urban resilience, equity, and sustainability. One avenue is investing in emerging technologies such as edge AI, which enables real-time decision-making without cloud dependency, improving latency and privacy. Seoul and London are leading edge AI deployment in traffic and surveillance systems. Similarly, 5G rollouts are enhancing bandwidth and connectivity for IoT and autonomous systems in cities like Singapore and Amsterdam.

Next up, inclusive governance frameworks are essential. Citizen participation through AI review boards, transparency audits, and open data initiatives are helping cities like Amsterdam and Helsinki balance efficiency with democratic accountability. Ethical compliance structures modeled on the OECD AI Principles and the EU AI Act are cited as necessary for long-term trust.

Third, research priorities must shift toward longitudinal, cross-sector studies that assess AI’s cumulative effects on sustainability and inclusion. The authors call for more studies measuring the environmental cost of AI infrastructure, such as training large models and operating extensive sensor networks. They also emphasize adaptive governance that can align rapidly evolving technology with dynamic urban needs.

The authors urge cities to integrate AI into disaster preparedness and climate adaptation. Predictive models for flooding and urban heat islands are already showing success in cities like Manhattan and Delhi. These tools offer scalable solutions for early warning systems and infrastructure resilience planning.