Tools
Tools
Adjei, Samson (2024) If he were motivated, how might Peter Zumthor engage with machine learning? Adapting new directions in computing to approaches in tradition oriented creative and architectural design processes. Doctoral thesis, London Metropolitian University.
Nearly all forms of architectural practice today are conducted using computers, instruments of a broader project which has progressed in a specific logical direction but also branched to a new direction through topical developments in machine learning or 'new computing'. Many architects embrace digital methods and advocate for the implacable objectivity afforded by computing in design; others, whose outlook regards culturally situated and historical contexts as foundational to the design process, view computing as antithetical to the craft of design. Both approaches are innovative in practice and rooted in expansive theoretical positions with rich interpretations of architecture, and both view this new computing with enthusiasm and concern in equal measure. However, it will be argued that each relies upon presuppositions, leading to charged debates which constrain interpretation of these new developments.
Situated in the design processes of architecture, this study positions machine learning at the forefront of the new direction in computing and in an unlikely relation to what will be outlined as a ‘traditional’ approach in architectural design, framing the question of how this approach might take up this direction. Contrasts will be made with another attitude in architectural practice to be outlined as the ‘digital’ approach, to understand established interpretations of conventional computing in architectural design. Yet, as this new direction has not yet achieved a level of incorporation in practice comparable with conventional computing, the study will look outside architecture to take inspiration from precedents in adjacent creative disciplines. Specifically, these will be precedents where machine learning is interpreted, transformed, or adapted for the aims of practices which share common ground with the traditional approaches. In so doing, the study will consider the question of whether a more nuanced, positive, form of discourse as a principle of a third approach, may be possible.
The first of three parts, Part A, will outline the philosophical foundations of both approaches, the logic of computing, and the underlying theoretical differences distinguishing machine learning from the conventional direction in computing. To situate this discourse, Peter Zumthor and Mario Carpo are positioned as two architectural scholars whose texts outline design processes, establish positions on computing in architecture, and reflect the two approaches whilst conveying broader theoretical contexts in relation to philosophical foundations. This will, however, highlight a limitation regarding architecture’s capacity to interpret this new direction in computing.
In addition to the still tentative presence of ML in architectural practice, Part B steps outside of the constraints of the discipline to consider instances in creative practices where adaptations of machine learning inspired reflection. These include writing, fine art, music, and cooking, returning periodically to chess as a touchstone—both a form of creative practice and a mediator between discourses. The research argues that comparative study can inform new insights by providing evidence for architectural reinterpretation. In Part C, we combine threads from the investigations to recontextualise theories of play, further inspired by chess, as a theoretical prism through which to illuminate a more meaningful conversation around machine learning regarding the attitudes of a traditional outlook in architecture
 |
View Item |