Urban AI
Published in

Urban AI

What artificial intelligence on social networks tells us about urban density.

Fig.1 — Agent-based modeling of a crowd of citizens interacting on an urban infrastructure network.

It is usually asserted that higher-density cities are more sustainable than low-density cities since they reduce significantly the length of public systems networks and the need for motorized transportation. But this assertion remains contested in urban planning theory: a direct correlation between urban density and air pollution has been established by NASA and urban developers admit unanimously that raising densities results in more expansive real estate prices, and impacts both buying power and the quality of living. A rising concern in Paris, for example, consists in addressing an answer to the fullness/emptiness observation along the suburbs-to-city-center rail system. The same line integrates hyper-dense urban areas with the highest prices of the world, with almost deserted villages that could offer much more space, gardens, and quality of life for an extra 20 minutes of transportation time. The temptation to invest in repopulating these abandoned settlements has never been so accurate since most city workers spend their days on phones and computers, and dream about vegetable gardens in their backyard. In my data science research, I have been interested in experiencing different computer simulations of the formation of public opinion and the influence of citizen engagement in decision-making for future cities. Empirical data was collected in three smart cities of very different cultural backgrounds: Taipei (Taiwan), Tel Aviv (Israel), and Tallinn (Estonia), and used to model artificial city contexts which reveal interesting findings between the density of social networks and the spread of sentiment in a population.

In Democracy Studio I am presenting in detail two agent-based models: a class of computational models from distributed artificial intelligence to simulate the collective dynamics of a crowd of autonomous intelligent agents interacting while assessing the unpredictable behavior at the scale of a complex adaptive system of mutual influences (see Fig. 1). The Stakeholder Engagement ABM is more specifically a simulation of the spread of two opposite opinions, which could be any positive or negative sentiment towards any random topic, in a network of undetermined citizens influenced by representative stakeholders of different categories: public sector, corporate company, startup business, academic sector, media industry, and civil society. Each of the stakeholders has been attributed a parametrization based on empirical data, while undetermined nodes representing the average population have been attributed mean values to the same parameters: engagement, influenceability, trustability, recovery capacity. An experience gain feature is added to take into consideration that citizens lose influenceability along with their opinion formation through multiplied stimulations. The interface of the model (see Fig. 2) allows different scenario testing by the interactive interface, and notably a variation of the network density which has significant consequences on the resulting output.

Fig. 2 — Interface of the Stakeholder Engagement agent-based model.

On the model interface, the second slider features the average node degree which in network theory directly matches with the network density. The higher the node degree, the more it is connected to other nodes, and thus the higher system density, defined as the number of connections in the network compared to the maximum number of connections possible. In this way, the formation of public opinion can be studied in a variety of population densities. Simulating a population of 100 citizens of a random social network across an artificial city, where one representative of each stakeholder category will be attributed a positive or a negative sentiment such as three of each will be distributed aside of 94 undetermined nodes, I will be comparing the results from sparse, medium, and high network densities. In sparse networks (average node degree of 1), a major proportion of neutral citizens remains while no clear majoritarian opinion emerges (see Fig. 3). In a medium level of density (average node degree of 3), a majoritarian opinion takes advantage of the whole network but allows the existence of neutral and opposite minorities (see Fig. 3), at the difference with higher density networks (average node degree of 5) where the majoritarian opinion take the full domination on the whole population, at an increased speed of more than twice faster (see Fig. 3) and gives no chance of subsistence to neutral or contrary opinions aside of the dominant one.

Fig. 3 — Formation of public opinion on varying network densities.

As we know that urban density boosts productivity and innovation, by easing interactions between similar-minded people sharing ideas and facilities in the same hubs, the stakeholder engagement ABM also shows that the resulting communities risk lacking diversity in their opinion and this might affect the performance of the system at the long-term. Indeed Internet activists already expressed their concerns about mass media issues: filter bubbles and echo chambers effects are underlying behind social networks’ algorithms. It affects our societies by creating intellectual isolation and radicalization of internet users who do not interact with people of different opinions. And the number of jokes from rural people on urban, as their equivalent mocking from urban people to the rural, are proving this effect of population density on mores is not a recent phenomenon of cities.

All the methods and tools to deploy the Stakeholder Engagement model on the cities of your interest are detailed in Democracy Studio book. It goes with online resources such as video tutorials and notebooks of code, accessible to programming newbies.

Further collaboration between Democracy Studio and UrbanAI is envisioned for the coming months, in the form of a working group on mutual topics of research. Follow us on social networks to stick around!

Urban AI is a Think Tank which federates a global ecosystem and a multidisciplinary community. Together, we propose ethical modes of governance and sustainable uses of urban Artificial Intelligences

Recommended from Medium

Why We Invested in Datomize

Stanford University Student and Innovator Inspires Young Innovators … in Sierra Leone (West Africa)

Umbrella Network Announces Partnership with VAIOT

Why I love mixing AI & AR ?

A Field Test; Can Chatbots Deliver The News?

Top Artificial Intelligence (AI) predictions for 2019

How the future of Voice is changing?

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store
Julien Carbonnell

Julien Carbonnell

Civic Technology and Smart-City. Data analysis, Machine Learning, Social Network Analysis, Computer Simulations. Project lead @ Democracy Studio

More from Medium

Sestina Bio — Harnessing the Power of Digital Biology for Synthetic Biology

A vision for a cooperative mobility data union

Agile Data Analysis upcoming courses on Urban Mobility Data Science