Terraforming Earth Lab 4: Naive AI for livable cities is part of ThingsCon, a joining of forces is sought between critical design practices around smart cities, the practice of IoT driven precision farming and unexplored potential of machine learning systems (framed by the ‘responsible AI’ discourse). The central question of the workshop is: (how) can the ecological development of cities be supported by machine learning systems?
Whether they label themselves as smart or not, cities around the world process more and more data, relating to mobility, housing prices, waste streams, air quality, crime, shopping behaviors et cetera. Most of this increased data processing is geared towards improving public (and private) safety, hygiene and traffic conditions, in general: to improve the economic and living conditions for human citizens.
Climate change with its rising temperatures, increased rainfall and increased atmospheric dynamism calls for a rapid adaptation of cities. ‘Greening’ is often mentioned in this context- increasing the number of trees. This would reduce temperatures, clean the air and would make cities more water resilient. But planting extra trees at a few busy and picturesque squares is not enough. The issues cannot be met with cosmetics only.
Cities need to be understood and developed in all their new and artificial ecologic richness and complexity. Urban ecology consists of interdependent events ranging from quick exchanges on the microscopic scale to the slow developments over large areas that take decades and include actors as diverse as microbiological soil life, humans and their pets, rodents, birds, trees, gardens, parks, pests, cars, warm water concrete, tarmac and different machines.
Policies that take a wider ecological understanding of cities as starting point for development project are all but absent. Also, the data that would be needed to develop such policies is not gathered sufficiently. If the data would be there, machine learning systems would come in useful to find patterns and correlations in the very diverse measurements on the artificial ecologies that are cities.
In this workshop participants focus on data gathering and data processing around the ecological characteristics of a city, and figure out how machine learning can assist in making cities more livable for the whole cocktail of 21st century society. What datasets and or sensor inputs can be used to tune machine learning systems towards favourable conditions for different urban ecology on different spatial and temporal scales? What interventions should the systems be able to make? How can they collaborate with the public services and the citizens?
This workshop is part of the research project Terraforming Earth.