Automatic Affinities proposes a framework for procedural design based on human-computer interaction. This system creates architectures of urban hybrids that serve as a new type of industrial space, intended for cohabitation of humans and machines. It is a system of design methods by the means of data curation, stochastic computation, and machine learning. The resulting architecture is a space of encounters and newly forged affinities that could not be otherwise achieved through conventional deterministic tools. In this framework, design becomes a series of orchestration: curatorial governance of sources, appropriation of processing resources, and editing of intermediary outcomes. In favor of chance encounters, predictive control of outcomes is deliberately suspended. Through data sampling, machine learning, and procedural assembling, the thesis proposes an interactive framework for human-machine collaboration in the design process, and simultaneously speculates a fantastically contemporary future in which men and machines think, make, and dwell in deliberate copresence.
A proposed building project is a hybrid campus of manufacturing, lab, and researcher-in-residence for General Motors’s autonomous vehicle division. After dispersion and degradation of its corporate bodies across America in the ethos of assembly-line style determinism, GM aspires to, in the heart of New York City, an advanced research institution for the probabilistic age. Situated on the fringe of Long Island City’s Anable Basin, the project mirrors the delirium of Midtown Manhattan with Roosevelt Island as a central axis. The project collects, processes, and assembles sampled fragments of both urban and corporate bodies of Manhattan and GM. The resulting architecture is a space of encounters and affinities made of self-affine parts coexisting as a hybrid urban body