“How could increasingly ubiquitous technologies like neural networks recognize the impacts of gentrification, economic imbalances and historical justice like redlining? Zeigler asked in his introduction. “Machine learning is indeed a construct: it’s not neutral, it’s not magic, and it ultimately reflects the subjectivities that are embedded in data sets. “

Each of the panelists and their affiliates have developed projects in recent years that take advantage of advances in advanced computing to understand our cities. Such advances, built on existing infrastructure and used to analyze already prevalent trends such as gentrification, nonetheless open up possibilities for as yet unrealized changes in urban space. For example, Rehm discussed student projects at SCI-Arc that used satellite imagery and machine learning to map potential sites for the construction of accessory housing units and solar panels; Hupalo described research using platforms such as Zillow and Redfin and their use to track speculative property sales in gentrifying Los Angeles neighborhoods; and Tedbury pointed out semblr, a project he created to imagine creative robotics-assisted construction projects using sustainable wood, creating flexible building structures that were explored by UK council estates as a way to better utilize public spaces.

Smith, the panelist whose work focuses on the city of Chicago, examines the ways in which changes in real estate markets are helping to explain the broader economic trends for the Windy City neighborhoods. A recent report from the Institute for Housing Studies examined the installation of 606, an elevated off-street path on a disused railway line, and how it fueled real estate speculation at its western end. Smith addressed the challenges of making important hypotheses about the future through simulation, noting how unpredictability has only grown in recent years, a crucial theme that has resonated in many years. numerous research projects of the panelist.

“Simulations and forecasts are all based on assumptions, and assumptions change,” Smith said. “All kinds of unpredictable things happen in the world, so I think if you’re transparent about the assumptions built into your model, it can be an interesting exercise. “