Towards an A.I. Design Architecture - Thesis Prep 2018 -

This thesis explores architectural design via Artificial Intelligence (AI). The goal is to generate architecture without direct human intervention. Today, computers are not yet smart enough to automate the architectural design process as it stands, not because architecture is too complex, but because architects have yet to become adept enough at communicating with AI. The issue here is not one of retaining control over an opaque discipline, but rather one of fundamental professional survival: if architects do not take control of automation technologies, other industries with more communicative flexibility will simply take over their historical roles. The choice is not between architecture and AI, but between architecture with architects and AI, and architecture without architects and AI.

Recognizing this, my aim is to redesign the architectural design process itself to more fully align with AI, shifting the creative burden from the production of architecture on a case-by-case basis to the production of new forms of overarching architectural thought that actively take advantage of AI.

There are two ways to generate form. One way is to designate forms to individual projects or parts of projects directly, and the other is to design the rules for an automated form-generating process (agent based or object-oriented programming). This thesis will explore the latter technique. This can allow a plethoric host of agents to participate in creating a form via a complex process that would far exceed the decision-making and organizational capacities of an individual designer. In this way, AI can design architecture that is more infrastructurally efficient and more responsive to a plurality of interests without sacrificing the aesthetic uniqueness that architecture traditionally prizes.