Relate, Relate, Relate: In the Age of Machine Learning

Abstract

Recognizing the impact of image-generating machine learning models on architectural discourse, this thesis offers a fresh perspective on the role of machine learning in conceptual relationships within architecture. The thesis explores ML's capacity to interrelate architecture beyond tradition lineage framework or categorization framework. Structured into three chapters, the first correlates projects from the "five on five" lecture series with large language and image-based models, forming a cloud of relationships. The second chapter delves into machine learning-aided design by relating projects and generating conceptual text. The final chapter investigates the challenge posed to museum design as the traditional architectural history framework is also challenged, proposing a museum embedded within a material reuse center. Through these explorations, the thesis uncovers ML's potential to contextualize and interconnect architecture, highlighting its significance beyond its prowess in generating realistic images and text.