Toward Urban Artificial Intelligence for Developing Justice-Oriented Smart Cities

Urban artificial intelligence (UAI) refers to the development and deployment of artificial intelligence (AI) technologies and solutions in urban settings such as energy management, environmental monitoring, public safety, transportation, and predictive maintenance. Urban research has shifted from a data-scarce to a data-rich environment in recent years. Big data and computational algorithms have become continually integrated into the built environment and within human’s daily lives, leading to a significant rise in digital twin research. Heterogeneous real-time data can be synthesized from a wide range of sources such as sensors and cameras connected to buildings, factories, green spaces, roads, sidewalks, and other urban elements. Defined as a virtual representation of a physical urban environment, digital twins can be employed to analyze, model, and simulate various aspects of urban phenomena in the fine scale. The UAI will take the source data from sensors, satellite imagery, and social media over space, time, and scale as inputs, and generate outputs that typically include predictions or simulations of how urban elements or systems will be affected by these inputs. Powered by UAI, the bi-directional flow of data between a digital twin and the physical urban environment further allows the digital twin to both reflect the current state of the city and make more informed decisions about how to optimize the urban operation (Ye et al. 2022). Furthermore, the revolution of computing power and information technology has blurred the boundary across disciplines and urban applications. Linked with immersive technologies such as virtual reality (VR) and augmented reality (AR), UAI helps people to visualize cities as they change by showing how planning and infrastructure design can alter the urban environment reflecting the diverse perspectives and needs of the community, either positively or negatively. Collaboration across disciplines and stakeholders is often essential for addressing complex urban problems. UAI has the potential to help bridge the silos within design, social, and engineering sciences as well as growing gaps between research and practice, through activities such as citizen science, community-based research, and participatory research. UAI is revolutionizing urban planning education and practice by providing new tools for planners to automate certain tasks and make informed decisions (Sanchez et al. 2022). This allows planners to focus on more creative aspects of their work and efficiently evaluate large amounts of design options, ensuring that the final delivery is an optimized solution. Simultaneously, UAI enables a collective understanding of existing urban conditions and demonstrates innovative capabilities for how to increase cyber, social, and physical resilience and efficacy beyond simple technology integration. However, UAI could possibly exacerbate existing socioeconomic inequities in the built environment if such technical strength is not designed and used by researchers and practitioners in an ethical and responsible manner. Hence, UAI needs to be used to promote knowledge co-production, which refers to the inclusive process in which knowledge is created, shared, and used through active collaboration among people with different backgrounds, experiences, and perspectives (Shrestha et al. 2017). Driven by UAI, knowledge co-production can engage a variety of relevant parties to identify research questions and co-design/ implement research projects through the process of data collection, analysis, and dissemination. In addition, UAI-based knowledge co-production can help cultivate the trust between researchers and the local residents, and increase confidence in its decision-making abilities, leading to a more comprehensive and inclusive understanding of the needs and priorities of the communities they serve. In terms of data shortage issue especially in lower-income or under-resourced communities, UAI techniques can be used to augment existing data with additional sources of information, such as public data sets or data from citizen science. UAI is highly affected by its training data and the adopted algorithms. As a result, it inherits the built-in biases embedded in these data and algorithms. The potential unfair or discriminatory outcomes may result in serious consequences for individuals or society if such biased UAI is deployed in decision-making processes. Hence, planners need to mitigate the bias by using representative training data, documenting the training process in an open-source manner, and consistently assessing the AI functions. We need to be transparent about the data employed to train UAI and the algorithms used to develop it. It is also important for developers to be aware of data ethics involved in the procedure of collecting, managing, and using sensitive data. It is vital to ensure that the data collection is with the consent of the involved individuals and that it is used in a responsible manner. In addition, the AI 1154002 JPEXXX10.1177/0739456X231154002Journal of Planning Education and ResearchEditorial editorial2023