Artificial intelligence in urban climate action

Climate change, alongside other interconnected crises, poses a major threat to livable cities. To help combat climate change, researchers and urban practitioners increasingly explore the potential of artificial intelligence (AI) as part of cities' mitigation efforts. In fact, some cities began deploying the first AI projects to help meet their climate targets. While research has mainly focused on the technical advancement of AI to help resolve climate change mitigation challenges, there is a need to better understand the ways through which cities have implemented AI in practice. Instead of focusing on the anticipated potential of AI, this dissertation makes a pioneering research contribution by investigating the use of AI in urban climate change mitigation as a socio-technical phenomenon. To this end, it combines systematically synthesized research evidence with empirical insights from real-world experience of urban practitioners, that is, a so-called down-to-earth approach. To address the overarching research question, `for what, how, and why is AI used in urban climate change mitigation efforts?', this dissertation draws on insights from three studies. First, evidence on the use of AI for urban climate change mitigation worldwide was synthesized through a systematic map of 2,300 peer-reviewed articles. This study identified 40 impact areas matched with mitigation options recommended by the Intergovernmental Panel on Climate Change. Research mainly focused on cities in East Asia, North America, and Europe as the main geographic research hubs. In the second study, practitioners' first-hand experiences of implementing AI for climate change mitigation were assessed across six European cities. Specifically, this study uncovered the roles of city administrations and key collaborators, as well as the prevailing challenges and solutions. Furthermore, it demonstrated how city administrations realized AI-for-climate projects based on three implementation arrangements. The final study of this dissertation investigated AI as a policy instrument in a single-city case study, scrutinizing the impact of socio-political dimensions of AI systems on the implementation of climate action. It showed how AI has incrementally reshaped the implementation of climate action across four mechanisms. Further, it revealed the technocratic, moral, and political logics that urban practitioners use to explain or oppose the integration of AI into their work. This dissertation enables better alignment between ML research, mitigation potential, and realities in practice. It marks the starting point for research to uncover city administrations' collaborative implementation of AI in urban climate change mitigation efforts and the influence of socio-political dimensions. Whether AI can make an effective contribution to urban climate change mitigation remains to be investigated. However, close collaboration between researchers and urban practitioners will be paramount to ensure technical advancement is aligned with urban realities and fosters livable cities.