Understanding our cities: a fusion of complex system science and AI.

By november 6, 2018 Algemeen No Comments

The fascinating thing about cities is that different aspects of them allow us to think about them in many different ways. At the level of urban infrastructure, cities certainly have features of machines, with vast constructed networks involved in transporting people, water, electricity, and waste.

At the level of the economy, cities resemble complex ecosystems, with companies and individuals filling specific niches and all living and working in a symbiotic dance. And at the level of growth and change, cities also feel like living, breathing, constantly growing and changing organisms.

But ultimately, the fact that a city has features of both a machine, a societal ecosystem, as well as a living thing means that a city is truly its own category: a novel type of socio-ecological-technological system that humans have made, and is perhaps one of our more incredible inventions.

When something is complicated, it is intricate but often lacks the dynamics that makes a system hard to understand. On the other hand, a complex system implies feedback, a sensitive dependence on the initial conditions, and emergent phenomena that are hard to predict.

As our cities’ systems grow and increasingly become interconnected, we are finding ourselves in a realm of the entanglement at the level of our cities. All cities face challenges, decision-making and accountability in a complex ‘system of systems’, of both traditional systems, such as critical infrastructure, as well as new ones resulting from emerging technologies, such as virtualization, sensor networks, etc. All aspects of a city’s life are complex combinations of events in both the real world (and physical space) and digital world (of cyberspace) and many transactions and interactions take place in or between both. Wherever they take place, the outcomes are certainly felt in the real world of a city’s stakeholders. For dynamical models to be realistic, they need to have accurate initial conditions, exact causality between systems variables and defined kinetics. The other issue with dynamical models of complex systems is the nonlinearity characteristic of complex systems. Because of the complex relationships between the variables in complex systems, the dynamics of the system quickly become nonlinear and complex.

A productive response involves looking at methodologies to understand living organisms or ecosystems. Since cities do resemble living things, at least in certain ways, perhaps we can use these approaches for living things and apply them to our own constructions, specifically our cities.

Obviously, ‘biological thinking’ can help us to better understand our urban environment, but we might also use ideas from physics to understand how innovation and productivity scale with the population of a city, network science to understand the many different diffuse networks that serve our cities, and even the quantitative social sciences to see how information spreads within an urban population.

We should be careful when we observe regularities in the global behavior of such systems: those regularities should not be taken as a clue that a formal, analytic explanation is lurking beneath the surface. The term complex system is used to describe precisely those cases where the global behavior of the system shows interesting regularities, and is not completely random, but where the nature of the interactions between the components is such that we would normally expect the consequences of those interactions to be beyond the reach of analytic solutions. Methods based on self-organization featuring emerging behaviors are very suitable for facing the complexity of the system.

These kind of systems are typically tackled by drawing inspiration from natural systems, where an intrinsic self-organization exist by which the globally intended behavior emerges out of local interaction of individuals. Given the intrinsic complexity of these applications then, new innovative techniques and tools for their analysis, design and deployment, are to be conceived. They  have to support the designer in controlling the emergence of adaptive behavior and in making qualitative and quantitate predictions about how the system works and about possible design errors.

There is a relationship between the study of complex systems, and the techniques of multi-agent system modeling. Multi-agent system modeling is about generative modeling of social processes that is computable in the mathematical sense. Urban components, actors are relatively simple components in the complex system of the city. Or more specifically, each component’s behavior is relatively simple compared to what the overall urban system is doing. A city as a whole is capable of engaging in complex behaviors. In contrast, no one single component, actor will have the impulse or knowledge to undertake such collective tasks on its own. It’s these collective behaviors that arise unexpectedly that are called ’emergent’ behaviors.

Complexity describes the behavior — it captures the available information, sensory capabilities, interaction dynamics and the range of possible actions a system can take. Complexity captures these degrees of freedom and the information available, while emergent phenomena are the actual behaviors, the occurrence or the appearance of those behaviors.

Emergence happens when the system has evolved to some critical point. In self-organized systems, critical states act as a kind of attractor. Once it reaches that critical state, the system seems to flip a switch and become resilient to future disruptions — the same disruptions that drove them to criticality in the first place. A collective then emerges, whose behavior as a whole is no longer correlated to the behavior of individual components. In this way, the system maintains its decentralized character, yet can act as a single entity. Thus, in tying the concepts to computational systems, the expression of any algorithms of these individual components must necessarily be simple, distributed and scalable. So while any system may begin with a simple set of components, under the right conditions, it will nevertheless be enough to generate a diverse range of differently-scaled systems, whether in nature or computing.

Ultimately, any sort of model that describes a complex system can be useful in providing insight into how cities operate, they must always be used with a certain amount of humility, in recognition of the complex reality that is a city.

The challenge of city intelligence is, to employ a biological analogy, more like genetic engineering than mechanical engineering, and part of the solution will require rewriting a city’s DNA.

Understanding and modelling cites, defined as the tightly intertwined social-economic-environmental system that humanity now inhabits, requires addressing human agency, system-level effects of networks and complex coevolutionary dynamics. Analyzing and understanding these dynamics sheds light on a coevolutionary view of urban dynamics in the Anthropocene including multiple development pathways, obstacles, dangerous domains and the sought-after safe and just space for humanity.

Theory and models of biogeophysical dynamics are well established, and our efforts developing  an adaptive network and flexible framework for modelling social-economic-environmental urban dynamics, regime shifts and transformations in an emergent and dynamic way, are offering interesting perspectives.  Dynamic prescription of scenarios, including phenomena such as social learning, segregation, norm and value change, and group dynamics such as coalition formation are very promising.

Developments in artificial intelligence relate to biological and the natural world. Many algorithms and systems have been developed inspired by systems found in nature. Examples are evolutionary algorithms, artificial neural networks computational immunity systems, bio-robotics, swarm intelligence or optimization algorithms based on colonies, hives, or flocks. These ideas transformed how we develop new technologies and solve problems. Bio-inspired solutions benefit from the fact that nature has already refined a lot of the steps to make them as efficient as possible. Thus, we will develop disruptive technologies by combining engineering and natures fine-tuned solutions.

Can we, based on this knowledge, experiences, develop intelligent models to visualize urban dynamic relations, feedbacks with the ability to process relevant (!) amounts of critical data? Urban modeling supported by artificial intelligence, with its ability to analyze scores of information from varied sources, can tease out the interactions between attributes and let us understand and predict the levers across the system we need to activate to enact change. Could the increasingly complex systems needed to manage the next generation of megacities become our first true artificial intelligence?  As we already have the models to understand and describe urban systems, the next step is to build artificial intelligence capabilities into our urban system modelling. A fusion of AI and complex system science will be vital to fully understand the urban web of life and maximize social-economic and ecological assets and values and catalyze a myriad of innovations. A real-time digital ‘dashboard’ for urban system management that would enable the monitoring, modelling probabilistic programming and an array of statistical machine-learning techniques. A decision support system for the management of these complex systems at a scale and speed never before possible. We have our urban system modelling and AI methods to do this. The challenge is to build something truly transformational, easy to use in real-time, open-access and data-dense, initiative uses machine learning and simulation modelling (urban design & architecture) to create a 3-4D living model of a (specific) city. This will require collaboration among partners and so we are building a platform (we still have some hurdles to take) that can bring the breakthrough: prototyping a Digital Urban Twin in its full form.

Leave a Reply


7 × zeven =