AI for climate action in cities: opportunities, challenges and what comes next

A new study, “Practical implementation of artificial intelligence for climate change mitigation in cities – priorities, collaborations and challenges”, published in Energy Research & Social Science, explores how city administrations in Europe are implementing AI to support climate change mitigation, what motivates them, and what obstacles they face.

The study’s main contribution lies in assessing first-hand implementation experiences, which is vital for understanding and shaping the deployment of AI-for-climate applications. The analysis is based on interviews with city staff and urban innovation experts across six European cities, sharing their experience from real-world AI-for-climate projects.

The study was led by Marie Josefine Hintz (Technische Universität Berlin, the Hertie School, and the Potsdam Institute for Climate Impact Research), and co-authored with Milena Gross Leuphana University in Lüneburg), Felix Creutzig (Potsdam Institute for Climate Impact Research and the University of Sussex), and Lynn H. Kaack (the Hertie School). Marie Josefine Hintz is a PhD student in the AI and Climate Technology policy group, led by Lynn Kaack, who is Assistant Professor of Computer Science and Public Policy at the Data Science Lab and the Centre for Sustainability.

Balancing priorities: Climate and costs

The research shows that city administrations are motivated to adopt AI-for-climate solutions not only to reduce CO2 emissions but also to lower costs. Rising energy prices and tight budgets make solutions that can, for example, improve energy efficiency particularly attractive. However, the study cautions that cities must carefully balance financial incentives with environmental priorities to ensure that climate impact remains the central goal.

Implementation with resources, skills and collaboration

Implementing AI for climate is resource intensive. City administrations need sufficient budgets, technical skills, and cross-departmental collaboration to develop, deploy and maintain AI applications. The results of the study suggest that city administrations take on multiple roles simultaneously, acting as developers, users and regulators to overcome technical and organisational challenges.

Collaboration is key, but the study found a surprising absence of big tech involvement. Instead, the European Commission plays a critical role, for instance, by providing seed funding for city projects. However, these funding mechanisms risk reinforcing existing inequalities by concentrating AI innovation in affluent regions.

In some cases, city administrations seek to develop internal capacities and retain a strong coordinating role, even if technical development is partially outsourced. In other cases, implementation relies on collaborative arrangements involving research institutions, innovation labs, start-ups, and publicly funded projects, often supported by national or EU-level programmes. A third pattern involves more traditional commissioning models, where external actors develop and deliver AI tools with limited internal involvement beyond project management.

Experimentation is not enough

The authors stress that ongoing experimentation alone cannot resolve the practical challenges of deploying AI in cities. Long-term investment in AI literacy, data governance, and ethical oversight is essential. Currently, many projects lack public participation, limiting opportunities to align AI solutions with residents’ needs. The study highlights the importance of integrating AI-for-climate initiatives into broader participatory processes to ensure transparency, trust and relevance.

Implications for city administrations

For city administrations exploring AI for climate, the research suggests three key steps. First, city staff should be upskilled, and positions should be created to manage AI-for-climate projects. Second, cities should be precise about the objectives of using AI, ensuring that solutions are appropriate for the climate problems they aim to solve. Third, experimentation should be accompanied by plans for maintenance and scaling, building on lessons learned and partnerships with local research institutions and innovation labs.

Future research directions

The study identifies four avenues for future research. One is to explore AI-for-climate implementation in different world regions to understand how context affects actor roles and challenges. Another is to develop tools for responsible AI governance that address ethical and environmental concerns. A third is to bring in social science perspectives to examine collaboration dynamics and power structures in AI deployment. Finally, more research is needed on AI applications for climate change adaptation, such as disaster response, which may face distinct implementation challenges.

This study provides an important first look at how AI can support climate action in cities, highlighting both opportunities and challenges. As AI systems continue to evolve, lessons from European cities can inform future policies, collaborations and investments to ensure AI delivers meaningful impact for climate mitigation and broader urban sustainability.

The full study can be read here.