Author Yuval Noah Harari (2016) and technologist Sebastian Thrun (2017) explain AI as a technology that allows us to do repetitive things in a faster and efficient way. AI is successful today because it can process a large volume of data much faster than a human being can. The availability of massive amounts of data combined with the capabilities of machine learning makes AI capable of re-writing its own code to outperform humans. AI has made its way in almost every field including urban planning. Urban planning is understood as the laws and policies that coordinate actions that may have an impact on the spatial form of the city. Planning is a coordination tool to manage contradicting interests and ensure efficiency and equity in the use of land and resources. Urban planners today are faced with the difficult task of balancing environmental and societal values, politics, technology and economics. They must cope with numerous challenges while trying to provide efficient infrastructure, housing, energy, safety, healthcare, education and a high-quality urban environment for residents in their city.
The development of AI in urban planning has increased in the last two decades. Wu and Silva (2010) provide a comprehensive overview on how AI is being used in urban analysis and modelling today. Tools developed from AI algorithms are integrated in existing analysis tools for example, within geographic information systems (GIS). GIS is a tool to visualize, analyze and interpret spatial data to understand relationships, patterns and trends in a better way. Integrating AI techniques increases the analysis capability of GIS to provide solutions for effective optimization of certain systems and processes. For example, land use allocation for new areas can be analyzed and mapped much faster by GIS when ES (expert system), which is a knowledge-based AI system that emulates expert decision making in a domain, is integrated within it. In transportation systems swarm intelligence (SI) which is an algorithm that optimizes decentralized, self-organizing systems is used to optimize traffic flows by providing alternative routes to the user. Using this reasoning method combinatorial optimization problems characterized by uncertainty for public transit schedules and ride-sharing can be overcome.
Wu and Silva (2010) also highlight that although different AI techniques are integrated within existing tools, approaches to integrate spatial and non-spatial factors in urban areas are scarce. This is because the AI techniques don’t have an integrated framework that merges spatial and non-spatial factors to develop within. Urban decision making is complex and requires the convergence of many actors. AI-based urban modeling is interdisciplinary relating to many different subjects such as engineering, architecture, geography, economics etc. Subjectivity of decisions that can be made by different actors and the limitations of the various interdisciplinary fields is reflected in the development of AI and can restrict its capabilities in urban modelling and decision making. Moreover, making these decisions raises big ethical and social questions. Consider the impact of self-driving cars on the urban landscape. The implementation of autonomous transportation system combined with SI could result in less traffic problems and increased safety. Such a transport system won’t require as much space as our highways currently take up. In this case, what will we use the space for? Climate mitigation measures? Or urban development? New public and community space? Or perhaps more cars? Some of these questions collide with sensitive ethical questions as well. Who should benefit from this? The local community? Or the companies who developed the autonomous vehicles?
This debate is a contentious one with no easy answers. Most of the choices that need to be made will be context based since all urban issues have different meanings in different geographical contexts. The non-spatial factors that influence these decisions play a very big role. Although trends in global urban issues are prevalent such as poverty or urbanization or the emergence of smart cities, the context of Beijing is different from Amsterdam or Riyadh or Mumbai. The inherent characteristics of local culture and attitudes, climate and governance are distinct and cannot be addressed in a ‘global’ solution.
The UN-Habitat (2015) points out that responding to urban issues through sustainable development is the biggest challenge lying ahead of us today. This is the accepted global norm for what planning should deliver. Explicitly or implicitly, urban planning emphasizes the need for sustainable development and provides a framework to tackle the uncertainties of urban issues. How will the development of AI affect this framework? Are we ready to address environmental and social questions that are becoming increasingly relevant through the development of AI? Today AI is successful in performing single domain tasks and is increasing the planner’s capability to make decisions about one issue such as land use change or site selection with more speed and precision. Not much progress has been made in integrating multi-domain tasks within the current capabilities of AI. This means that even AI as it exists today, doesn’t have what it takes to integrate data to make decisions and plan cities better than human beings. This could change in the future as an increasing amount of applications are being developed and tested. Knowledge-based intelligent systems could potentially change the way in which we perceive and use data in urban decision making. Planners are already conflicted between balancing technology, politics, economics, environment and societal values. They need to position themselves between professionalism and politics. Will this dilemma be amplified or reduced with the progress of AI? Maybe it is too soon to answer this question. Perhaps the next generation of urban planners will need to rely on their innate emotions and sense of empathy, authenticity, integrity, imagination, intuition and above all honesty to guide their decisions. Planners also need to rely on their collaborative capacity to solve problems since no individual can visualize or comprehend the complexity of urban systems. They form a small part of the multi-layered web that no single person really understands completely. Going ahead we need to ask ourselves if we are adapting our skillset and qualities to be capable of navigating through an even more complex urban environment to deliver a sustainable future for our globe.
Harari, Y. N. (2016). Homo Deus: A Breif History of Tomorrow: Harvill Secker.
Ning Wu, E. A. S. (2010). Artificial Intelligence: Solutions for Urban Land Dynamics: A review. Journal of Planning Literature, 24(3), 246-265. doi:10.1177/0885412210361571
Thrun, S. 2017. TED talk with Chris Anderson – What AI is and isn’t. Available from: https://www.ted.com/talks/sebastian_thrun_and_chris_anderson_the_new_generation_of_computers_is_programming_itself/up-next
UN-Habitat, U. N. H. S. P. (2015). Urban Solutions. Retrieved from Nairobi.