Unboxed City

Critical explorations of [ai] and cities



RoleDesigner, Researcher, Data Analysis, Website Implementation.

Team and Partners
  • MIT’s Leventhal Center for Advanced Urbanism
Project Description“Unboxed City: Critical Explorations of [AI] and Cities” immerses us in the workings of generative AI algorithms by placing us inside a literal black box. This large black box symbolizes the concept of black-box algorithms, first introduced in cybernetics, which refers to the often opaque nature of understanding how these systems function. At the center of this box, a totem with four large screens displays multi-directional videos that illustrate the process of neural networks—collecting, encoding, and decoding data to transform it into artificial intelligence.

The narrative exposes the biases and limitations inherent in shaping cities through AI. These algorithms rely on accessible datasets such as satellite imagery, street-level photos, demographic profiles, and climate information, but they lack the nuances that truly define a city. The black reflective walls inside the box project a continuous loop of real and unreal elements, simulating the experience of being inside an algorithm. This infinite reflectivity serves as a reminder that generative AI creates impressions of reality, not reality itself.

The experience critiques the biases embedded in AI, highlighting how incomplete, unknown, unequal, and non-inclusive datasets often leave much of the city out of focus. It underscores the necessity of incorporating human input and community perspectives into AI models to ensure they reflect the diverse needs, desires, and dreams of citizens. Ultimately, the installation calls for the development of AI systems that are cooperative with human intelligence and not dominated by power. It advocates for AI that represents the diversity and complexity of the world beyond algorithmic constraints.
Websiteunboxedcity.mit.edu


@media (max-width: 768px) { .column-set { height: 20vh; } }




Alberto Meouchi
2024