By Sarah Popelka
Sarah Popelka, Laura Narvaez Zertuche, and Hubert Beroche recently released the Urban AI Guide, a practical handbook for urban technologists, which contains an anatomy of an urban artificial intelligence system, as well as a primer on responsible implementation. This guide arose from conversations with city leaders, who were confronted with new technologies (like artificial intelligence) as a means of solving complex urban problems, but who felt they lacked the background knowledge to properly engage with and evaluate the proposed solutions. In some instances, this knowledge gap produced a barrier to project implementation or led to unintended project outcomes.
More and more, urban practitioners seek informational and technological solutions to the enormous challenges that cities face. The proliferation of big data, coupled with massive advancements in the field of artificial intelligence, has propelled municipalities into the tech development space, often with limited resources and knowledge. The result, so-called “urban artificial intelligence” (referring to any system that incorporates data derived from the urban environment, which is then processed by algorithms, the result of which has useful applications in the socio-spatial nexus of the city), has three distinguishing features which necessitate a deeper understanding of such systems: the complexity of the city, the specific policy contexts, in which urban artificial intelligence operates, and the hybridity of urban artificial intelligence. Engaging with artificial intelligence in an urban context requires an intersectional approach combining the technological, the urban, and the human.
To inspire such an approach, the Urban AI Guide dissects an urban AI system, exposing its core components. In doing so, it presents a framework for understanding the steps and considerations that go into implementing an artificial intelligence project in an urban context. Without urban infrastructures, i.e. the intertwined layers of the built and physical environment, flows, governance, and people, there would be no city. The multiplicity of stakeholders operating in metropolitan areas requires the identification and consideration of many actors. Sensors and data collection infrastructures allow for the capture of information emitted and transmitted by urban systems and processes. These can range from small, community-led efforts to massive city-wide sensor networks. Network infrastructures transfer information to data storage infrastructures, which allow for the organization and retrieval of data. Data privacy, security, and size drive decisions around the storage of data. Due to the unpredictability of a real-world environment, it is rare that incoming data will have an adequate format and no errors; data processing facilitates the assessment and cleaning of raw data. Data visualization aids analysts in better identifying patterns and communicating information. Artificial intelligence/ machine learning, can be utilized to enhance the completion of any of these steps, as well as for the implementation of analysis and predictions. Finally, decision-making/ adaptation transforms the insights gained in the process into action, in a continuously evaluated process.
Through the use of detailed case studies from three different cities, the Urban AI Guide demonstrates what urban artificial intelligence can look like, practically, and how it can be applied in different disciplines and urban contexts. A water meter lifecycle project in Winnipeg, Canada outlines the considerations necessary to build out an internal artificial intelligence team and shows how municipalities can take a phased approach to project implementation while they build capacity. A curb digitization for planning and asset management project in Los Angeles, United States highlights the role that a public private partnership can play in expediting project timelines, and underscores the importance of scaling the solution to the specific problem and urban context. An air quality sensing project in Vilnius, Lithuania shows how platform/ package-based AI service providers can provide AI capabilities with little overhead on the part of the city. Future iterations of this guide will further expand the geographic and sectoral scope of case studies presented.
The literature review, urban AI anatomy, and case studies all fundamentally highlight the role of human beings in implementing, overseeing, and evaluating urban AI projects. Since urban AI projects utilize human-based data, the physical components of urban AI systems interact with the human-built environment, and the decision-making that amounts from an urban AI project affects humans, people must remain actively involved in each component of the project. In this sense, communication plays a key role: engaging with the community at key moments, building intelligibility mechanisms such that algorithms do not exist as black boxes, and conveying an understanding of the system, holistically, to residents and decision-makers alike. Effective and responsible change does not arise from reacting to model outputs at face value. Instead, urban practitioners ought to remain actively involved in crafting, evaluating, and re-designing urban AI processes. Urban artificial intelligence holds its greatest value not as a means of making decisions, but as a means of informing them.
To read the guide in full and learn more about the project, please visit: https://urbanai.fr/our-works/urban-ai-guide/.
By Sarah Popelka, Contributor and Advisor at Urban AI
