Neural Architecture – Design and Artificial Intelligence
Neural architecture is the field of architecture that is primarily preoccupied with interrogating the emergent field of artificial neural networks (ANNs) as a design method.
This lecture presents an attempt to utilise Deep Learning (DL) and Machine Learning (ML) to capture the salient features of existing architecture in order to interrogate these data for their underlying architectural qualities.
What is meant by underlying architectural qualities? The rational explanation would include aspects such as spatial layout, sectional distribution of volumes, the dialog with its environment, the volumetric balance, the material qualities of the design, the structural properties, etc. All of these can be explored with the help of ML processes. However, this lecture maintains that architecture is more than just an assemblage of rational properties. This might explain the obsession with neural art, which represents an excellent mirror of our contemporary age, particularly regarding our shared agency with quasi-intelligent machines and their observation of the world.
Can architecture do the same? Can neural networks help interrogate the latent layers within the geological deposits of the history of architecture and then assemble those found features into hitherto unseen designs?
Matias del Campo is a registered architect, designer, and educator. He is an Associate Professor at the Taubman School of Architecture and Urban Planning, University of Michigan, Director of the AR2IL, The Architecture and Artificial Intelligence Laboratory (AR2IL) at UoM, and an affiliate faculty member of Michigan Robotics, Computer Science, and Data Science. Matias is also the co-founder of the architecture practice SPAN. The practice gained wide recognition for the design of the Austrian Pavilion at the 2010 Shanghai World Expo and, more recently, for the Robot Garden at the Ford Robotics Building. SPAN’s work was featured in the Venice Architecture Biennale in 2012 and 2021, at ArchiLab in 2013, and at the Architecture Biennale in Vienna and Buenos Aires in 2019. Solo shows include «Formations» (MAK, Vienna) and «Sublime Bodies» (Fab Union, Shanghai). SPAN’s work is in the permanent collection of the FRAC, MAK, Benetton Collection, Albertina, Pinakothek Munich, and several private collections.
His publishing work includes two editions of AD – Evoking through Design and Machine Hallucinations (co-edited with Neil Leach) as well as the books Neural Architecture – Design and Artificial Intelligence (ORO Editions 2022) and Sublime Bodies (co-authored with Sandra Manninger, Tongji Press 2017).