What Architecture as a collective strives for is hard to define. However, a common thread of connection can be understood as an innate search for pattern, or to put it in a more corny term, the truth, be it the deadpan treatment of historic and political artifacts manifested into physical environments or a longing for utopia. Artificial intelligence has given the world a tool to grasp a sense of pattern. Given the right objectives and ask the right questions, what is fundamentally just a mathematical framework can model our world to high level of precision. Artificial Intelligence has already been an extremely useful tool in fields such as mathematics, physics, economics and psychology. As a profession that seeks to understand how human beings interact in and with space, artificial intelligence seems to be a critical tool waiting to blossom.
Previously, researches have been done with various generative methods for architecture. Some notable examples are urban block design automation carried out by researchers at MIT media lab, floor plan generation algorithm carried out by Ruizhen Hu et al. and Stanislas Chaillou. This project is a continuation of the study of applying artificial intelligence in architecture. To narrow down the problem, we are focusing on improving layout generation models by improving and changing GAN architecture. This is meant to be a proof of concept of the infinite possible ways data-driven methods can be applied in architecture and adjacent domains. Perhaps there’s are quantitative metrics out there able to help us understand what makes a built-environment and what it means to be a good piece of architecture.