Artificial Intelligence and the Smart City

More than half of the world’s population, some 4 billion people, live in cities. By 2050, this proportion will be closer to two-thirds. Like it or not, humanity is increasingly an urban species. Cities and most 21st century problems, from climate change to the global housing crisis, are at least partially urban at their core. The same can be true for the topic of artificial intelligence. The technological buzz world of the hour, AI has received an immense amount of attention from journalists, politicians and academics alike. While the ethics of AI more broadly have been a serious subject of debate, many have failed to find the fundamental connection between the development of AI and the city in the 21st century. 

Less than a decade ago, researchers shocked the world when Google’s AlphaGo AI beat champion Lee Sedol in a five-round match of Go, the abstract strategy board game. Today, AI is a concrete reality in many cities, woven into their very social and spatial structure. AI is often implemented as part of “Smart City” initiatives, a rather vague term which refers to cities seeking gains in efficiency and continuing development through the use of information and communication technologies. In some cities, AI is at work piloting autonomous cars. In others, it monitors the flows of traffic and adjusts stop lights to assure efficiency. Current iterations of urban AI are generally narrow ones, characterized by artificial intelligence capable of performing one, specific task or function rather than dealing with any general task at hand. Nevertheless, urban AI is having a marked and oftentimes unequal impact on the cities in which it is at work. 

Many of the ethical issues raised regarding the current iterations of urban AI are much the same as the broader critiques of the technology. Primarily, fears regarding privacy abound. Before the widespread adoption of artificial intelligence in cities, researchers and activists were already concerned with the vast amount of information gathered regarding the inhabitants of the city. Whether sourced from device-generated location tracking or the large networks of surveillance cameras which often stretch across the world’s most populous cities, companies and agencies have access to vast amounts of information about the city’s citizens, including intimate personal identification data. Today, with powerful AI available, such data can be quickly sifted through and analyzed to build knowledge and inform action. The enhanced data sorting and analysis tools offered by AI and the increased appetite for such knowledge reflect a further erosion of the right to privacy. 

Another ethical question regarding the current iteration of urban AI represents a revival of a classic question regarding morals and ethicsIn the case of Masdar City, the archetypical urban experiment constructed from scratch in the desert of the UAE, plans for automatic transport have quickly given way to those of autonomous transport. While automatic transport was planned as a sort of shuttle or metro system, the newest initiative is that of autonomous, or self-driving (AI) cars. Unconfined to a single track or route, such cars take in information from their surroundings through sensors, analyze the data to form knowledge, and then act upon such knowledge. When autonomous vehicles move through the streets of a city, space is being shared between human and nonhuman intelligence. Imagine that an autonomous car must choose between turning and hitting one young woman or staying straight and smashing into three old women. Such a scenario, a twist on the classic “trolley problem”, shows the limits to the knowledge possessed by AI. AI must be trained regarding ethics and the so-called right thing to do. However, different individuals, organizations, and datasets will present distinct opinions on this subject. How can we know how an AI will react when the information from which it learns cannot present a compelling conclusion? 

While AI is at work in the city today, much of it is confined to the spaces subject to “smart city” initiatives. Such initiatives, while popular and widespread, are often only available to a particular subset of cities, generally those with the means to invest in such technologies. Moreover, once implemented in a city, they are often done so in a piecemeal fashion which can serve to reinforce pre-existing power imbalances within the city – AI tends to be implemented in wealthier neighborhoods. Nevertheless, the use of AI in cities will only grow as the 21st century progresses. Far from a niche tool on the cutting edge of technology, AI will, by all predictions, become ubiquitous across the urban centers of the world. Not will the geographic extent of AI spread, but so will the scope.

Urban geography perspectives are particularly helpful for understanding the potential future of urban AI. Urban geography, the sub-discipline of human geography concerned with studying the city and city life, conceives of the city as a socio-spatial process. Boarding on academense, the phrase socio-spatial process actually tells us very interesting and important things about how geographers understand the city. Firstly, socio-spatial means that geographers understand the city to be composed not only of concrete and steel (spatial) but of relationships between people (socio). Urban geographers emphasize how material elements of the city and the human ones interact and influence each other. Moreover, process refers to the idea of the city as constantly changing, shaped by struggles between differently powered groups over history. Therefore, urban geography sees the city as a twinned material/human system which changes and develops over time. 

One urban geographer particularly interested in this issue, Federico Cugurllo, examines the potential ramifications of  future urban AI technology, namely that of AI-driven urban governance and planning. According to Cugurullo, AI governance and planning in cities could undermine our fundamental understanding of urbanity. Returning to the socio-spatial process idea, when data and artificial intelligence run the show through planning and governance, the” socio” element of the city is weakened. Instead of a living and breathing network of humans with dreams and aspirations, the city is reduced to its brutish materiality, tons of concrete, steel and glass adorned by heaps of data. Such a perspective is like reducing a human body, with its muscley, fleshy materiality, to a mere dry skeleton. 

Moreover, the idea of AI-driven urban planning is troubling to the ethics-minded because of the importance of values and ethics in urban planning itself. How will AI develop understandings of the ideal city? Looking to real-life examples to learn about urban planning would entail repeating the same mistakes of history. Tackling issues such as urban planning and governance involves generating subjective ideas about how we ought to live together. Whereas AI is often implemented in technocratic, top-down smart city initiatives, the process by which we determine how best to live together should be a grass-roots, participatory one. AI is only as good as the information it is fed, as they say in statistics: garbage in, garbage out. 

All and all, while technological developments will certainly be a part of the solution to many 21st century issues, technology is not a panacea for our modern woes. The issues with (urban) life require an understanding of the sociality of the city and a deep, participatory and democratic discussion regarding how we wish to live together as we move further into an unpredictable and rapidly changing world.