AI That Reads Cities: Between Design Tool and Surveillance Concerns

According to AI TIMES, the MIT Senseable City Lab said on September 24 that visual AI could become a tool for reading cities more precisely, but that it could also intensify surveillance and bias risks. The research team compiled these potential uses in a book, including applications such as analyzing video from 331 New York City traffic cameras to closely track urban flows and environmental conditions. The key point is that images are being transformed from simple records into data used for urban planning and policy decisions. The lab said visual AI can capture street-level perceptions of greenery, ways public spaces are used, and even the local character of indoor spaces, which satellite images can easily miss. However, it noted that results can vary depending on what the system was trained to see, which can reproduce existing social prejudice, and that large-scale video collection could deepen concerns over privacy violations. In the end, it raises the question of how to design not just the technology’s performance, but also the standards for observation, interpretation, and use, in order to strike a balance between smart city benefits and citizens’ rights.