Unboxed City

at the Norman B. Leventhal Center for Advanced Urbanism
East Lobby, MIT Media Lab, 75 Amherst St, Cambridge, MA 02139

from March 8, 2024 to May 6, 2024

Unboxed City

critical explorations of ai and cities

“Unboxed City: critical explorations of [ai] and cities” teaches us how generative AI algorithms work by literally being immersed within one. The large black-box references the notion of black-box algorithms, a concept first introduced in cybernetics referring to the inability to understand how systems/algorithms work. A totem in the center box holds four large screens playing multi-directional videos illustrating the process of neural networks—collecting, encoding, and decoding data, thereby transforming it into artificial intelligence.

The narrative reveals the biases, and limitations in shaping cities through AI which are built on available datasets—from satellite imagery to street-level photos, demographic profiles, and climate information—they are assembled on what is accessible, not the nuances of what makes a city. The black reflective walls of the box’s interior project real and unreal elements layered in an infinite loop, creating the sensation of what it might feel like to be inside an algorithm. The infinite reflectivity reminds visitors that what is created through generative AI is not real but rather impressions of reality.

The experience critiques the bias embedded in AI highlighting incomplete, unknown, unequal, and non-inclusive datasets leave much of the city out of focus. It is clear there is a need to incorporate human input and community perspectives into models to ensure they reflect citizens' diverse needs, desires, and dreams. Ultimately, there is a call to action to generate AI systems that are not dominated by power but rather are cooperative with human intelligence—representing the diversity and complexity of seeing the world outside of the algorithm.