How to Put AI at the Service of People in Cities, and Not Otherwise: in Conversation with Prof. Flavio S Correa da Silva

How to Put AI at the Service of People in Cities, and Not Otherwise: in Conversation with Prof. Flavio S Correa da Silva

Prof. Flavio S Correa da Silva is a researcher in Artificial Intelligence, with over 30 years of experience in theoretical advancements of AI, as well as the development of AI applications for problem-solving, with a particular interest in human-centric AI and applications related to urban mobility, healthcare, wellbeing, and the ageing society. After completing his PhD in 1992 at the University of Edinburgh (UK), he returned to Brazil to become a faculty member at the University of Sao Paulo (USP), where he created the Laboratory of Artificial Intelligence, Logics and Formal Methods and the Laboratory of Interactivity and Digital Environments. He is currently Associate Professor at USP and Honorary Research Fellow at the University of Aberdeen (UK), having held the position of short-term visiting scholar at the University of Milano-Bicocca (Italy) in 2018 and 2019. He has published over 100 articles in peer-reviewed journals and conferences and written eight academic books.

Q: There seems to be a growing interest in building Intelligent Cities. From your perspective, what could truly bring intelligence to cities?

FS: The concept of Intelligent (or Smart) Cities (IC) is usually tied to the use of computational technologies to improve the efficiency of urban services. Even though this perspective of IC can be traced back to the late 1960s, it is commonly assumed to have gained prominence in the 21st century. It is rooted in the notion of cities as machines, in which citizens are components in a large mechanism. Frequently, it induces the search for solutions to specific and specialised problems (such as over-consumption of natural resources, environmental pollution, and efficiency in the delivery of public services), without challenging relevant sociotechnical imaginaries (e.g., What causes the consumption of natural resources and environmental pollution? What services are perceived by citizens as most relevant and based on what presuppositions?). In contrast, Human-centric Urban Design (HCUD) is based on the principle that cities (and related technologies) exist to serve people, hence citizens are not components of urban systems, rather they constitute the raison d’être of cities. HCUD has its origin connected to the works of Jane Jacobs and particularly of Jan Gehl and has proven to be very effective at getting to the root of problems and issues to be solved, by giving priority to human and environmental needs and valuing local knowledge. In my view, this is the right approach to designing and building truly intelligent cities.

Q: How can AI be useful to design truly intelligent cities?

FS: The goal of AI is to build systems that can sense, reason about, and act upon the states of complex environments, based on computations that enable complex behaviour to the point of deserving to be classified as intelligent. Since its origin in the mid-1950s, research in AI has been structured in two lines: systems to replace human agency and systems to augment human capabilities, more recently coined Human-centric AI (HCAI). The former aligns well with the mechanistic approach to the analysis and design of complex systems, whereas the latter aligns with the human-centric approach. HCAI can provide urban designers, planners, and managers with effective tools to preserve, support, enrich, and evolve local knowledge. To name a few possibilities, route-finding systems for pedestrians can take into consideration the health and wellbeing attributes of all citizens on a street, au pair with the optimisation of distance, speed, and safety of individual pedestrians, and generative design can be enriched by large scale, location-aware empirical data to provide designers with innovative design alternatives.

Q: There has been growing debate about ensuring ethical behaviour of AI-based systems. How does this issue impact the use of AI in the design of truly intelligent cities?

FS: Broadly speaking, Ethics is about the identification of attitudes that can take humans – and Humanity as a whole – to a life worth living. The field of Normative Ethics is commonly divided into three main approaches, namely (1) Ethics of Virtues, based on the view that the development and practice of virtue (generosity, courage, compassion etc.) forms the basis of a life worth living; (2) Ethics of Duties, based on the view that ethical behaviour must obey socially established duties and obligations; and (3) Ethics of Consequences, based on the view that what is right or wrong must be judged by a balance between positive and negative consequences of actions. Monitoring and assessment of alignment with respect to norms and duties, as well as assessment of short-term consequences of actions, are relatively easy to implement and automate in computational systems. However, they work best to identify misalignment with respect to norms and expected positive balance of consequences. Nevertheless, positive outcomes of intelligent systems are more naturally connected to the individual and collective nurturing of virtues. Recently, I have proposed the development of Positive AI, specifically aiming at this ethical issue with respect to HCAI to support truly intelligent cities. In my view, this approach to the development of AI can lead to tangible quality metrics for systems design to support the design of truly intelligent and human-centric cities.