A citymaker’s guide to… Artificial Intelligence : What happens when cities go digital?

Rethinking AI, data and public value in urban areas

Laura Valdés, Head of Policy, 2025

To download : metropolis_layout_ai_web.pdf (1.1 MiB)

The Stakes: Why AI Matters for Cities

The Urgency of Urban AI

Cities are evolving into data-driven ecosystems where AI acts as both a catalyst for innovation and a potential disruptor of social equity. The integration of digital twins (virtual replicas of physical systems), predictive analytics, and immersive platforms into urban infrastructure promises to revolutionize how cities manage resources, respond to crises, and engage citizens. For instance, Hangzhou’s City Brain, developed by Alibaba, uses real-time data to optimize traffic flows, reducing emergency response times by nearly 50% (UN-Habitat, 2024, p. 12). Yet, this same infrastructure can enable mass surveillance, raising ethical concerns about privacy and consent (Wired, 2018).

However, the rapid adoption of AI outpaces regulatory frameworks, leaving cities vulnerable to bias, disinformation, and environmental strain. For example, AI systems trained on incomplete datasets have been shown to reinforce discrimination against marginalized groups, such as women and ethnic minorities (UNESCO, 2021, p. 8). Meanwhile, the energy demands of AI models—particularly in data centers—pose significant sustainability challenges, with some estimates suggesting that training a single AI model can emit as much carbon as five cars over their lifetimes (MIT News, 2025).

Key Risks:

• Inequality and Bias: AI systems can perpetuate discrimination when trained on biased data. For example, Spain’s VioGén system, designed to assess domestic violence risk, failed to identify women who were later killed by their abusers, revealing fatal flaws in its algorithmic design (The New York Times, 2025, p. 15).

• Disinformation: AI-generated deepfakes and misinformation threaten public trust, particularly during elections or emergencies. The World Economic Forum ranks disinformation as the most urgent short-term global threat (Global Risks Report 2025, p. 5).

• Over-Reliance on Automation: AI lacks the contextual understanding to navigate the complexities of urban life. For instance, predictive models in Los Angeles County helped prevent homelessness for over 700 individuals but required human oversight to address systemic biases in housing allocation (CalMatters, 2024; Vox, 2024).

• Privacy and Sovereignty: Closed AI models, such as those used in Metaverse Seoul, raise concerns about data exploitation and vendor lock-in, as residents interact with virtual administrative services without full transparency on how their data is used (Seoul Metropolitan Government, 2023).

• Environmental Cost: The carbon footprint of AI training and deployment is substantial. For example, data centers consumed  1% of global electricity in 2022, a figure expected to rise as cities expand their digital infrastructure (MIT Technology Review, 2022).

Cities stand at a crossroads, where the promise of AI—smarter infrastructure, faster responses, and deeper civic engagement—collides with the reality of its risks: bias, surveillance, and environmental degradation. The challenge is not just to adopt these tools, but to wield them with wisdom, ensuring they serve all residents, not just the privileged few.

Challenges: Lessons from the Field

Rethinking Participation in the Digital Public Square

Digital platforms are transforming civic engagement, but their success depends on inclusivity and adaptability.

• Madrid (Spain): The Decide Madrid platform, supported by the AI assistant CLARA, has involved over 60,000 residents in shaping local policies. However, engagement remains uneven across age, gender, and income groups, highlighting the limits of digital participation (City of Madrid, 2025).

• Los Angeles (USA): After the devastating 2025 wildfires, the Engaged California platform was repurposed to involve residents—particularly those most affected—in shaping recovery plans. This community-first approach ensured that marginalized voices were prioritized in rebuilding efforts (Engaged California, 2025).

• Cape Town (South Africa): The Innovation Hackathon invited young coders to design solutions for urban challenges. The winning project, Citi Builder, is a low-cost platform improving access to city services and employment opportunities, demonstrating the value of local talent in creating accessible tools (City of Cape Town, 2024).

• Seoul (South Korea): Metaverse Seoul allows residents to access administrative services, submit tax queries, and even visit the mayor’s office virtually. Despite its innovation, the platform has struggled to maintain public interest, raising questions about the long-term viability of virtual engagement (Seoul Metropolitan Government, 2023).

Bringing Public Services Closer to People

AI can improve service delivery, but oversight and human judgment remain critical.

• Buenos Aires (Argentina): The BA INFINITE platform uses metaverse-based learning to teach history and science. Students can explore virtual environments, such as the city in 1810 or a planetarium, blending education with digital literacy (City of Buenos Aires, p. 10).

• Los Angeles County (USA): A predictive model analyzing data from hospital visits to food aid usage identified individuals at risk of homelessness. Since 2021, the system has helped 700+ people retain housing, but its effectiveness depends on sustained investment in affordable housing (CalMatters, 2024; Vox, 2024).

• Algorithmic Bans (USA): Cities like Minneapolis, San Francisco, and Philadelphia have banned algorithmic tools used to inflate rental prices, recognizing that unchecked automation can harm public interest (Stateline, 2025).

Improving Metropolitan Planning and Disaster Preparedness

AI is reshaping urban planning, but ethical governance is essential.

• Hangzhou (China): The City Brain initiative, developed by Alibaba, optimizes traffic and emergency responses but has faced criticism for enabling mass surveillance (Wired, 2018).

• Barcelona (Spain): The City OS platform promotes ethical digital governance through public algorithm registries and mandatory audits, ensuring transparency in data usage (City of Barcelona, 2021).

• Istanbul (Türkiye): A digital twin simulates earthquakes to test emergency responses, but with 600,000 homes at risk of collapse, technology alone is insufficient without investment in safe housing (Turkish Minute, 2025).

• California (USA): The AI-powered wildfire detection system showed early promise but failed to prevent the 2025 wildfires, which caused 30 deaths and $250 billion in damages (TIME, 2023; Los Angeles Times, 2025).

From the virtual streets of Metaverse Seoul to the wildfire-ravaged hills of California, cities are testing the limits of AI’s potential. Each example reveals a truth: technology can save lives or erode trust, optimize systems or entrench inequality. The difference lies not in the tools themselves, but in the values that guide their use.

Recommendations: A Roadmap for Ethical AI in Cities

Critical Questions for Policymakers

To ensure AI serves the public good, city leaders must address the following:

1. Co-Design: Are digital platforms co-created with residents, including migrants, disabled individuals, and informal settlement dwellers?

2. Community Stewardship: Is the city investing in local governance models to ensure meaningful, ongoing participation?

3. Safeguards: Are bias audits, algorithm registries, and ethical frameworks in place to prevent harm?

4. Long-Term Governance: How will digital systems outlast election cycles and align with social and environmental goals?

5. Capacity Building: Does the city have the skills and staff to manage and adapt AI systems over time?

6. Data Openness: Is data standardized, shared, and improved across departments and jurisdictions?

Actionable Steps

• Transparency: Mandate public algorithm registries (e.g., Barcelona’s City OS).

• Accountability: Regularly audit predictive tools (e.g., revising Spain’s VioGén system).

• Inclusion: Prioritize community-led initiatives (e.g., Cape Town’s Citi Builder).

• Sustainability: Assess the environmental impact of AI systems (e.g., MIT’s research on AI’s carbon footprint).

The path forward demands more than technological sophistication; it requires a commitment to equity, a willingness to question, and the humility to adapt. Cities must ask not just what AI can do, but for whom—and at what cost. The future of urban AI is not predetermined; it is a choice, and the time to choose wisely is now.

Sources

To go further