Semiramis - Vertical Garden - Automatic AI generation of designs

Usage of AI method described in the paper "Augmented Intelligence for Architectural Design with Conditional Autoencoders: Semiramis", allows architects to request some desired properties and obtain multiple predictions of designs that fulfill them. The results provided by the model can oftentimes go beyond what the user expected and thus augment human's understanding of the design task and stimulate design exploration.
This methodology provides much more flexibility than standard manual exploration approaches, which require a tedious manipulation of the input parameters, as depicted in the video: • Semiramis - Vertical G...
This type of human-in-the-loop approach will enable a richer and more exhaustive exploration of the design space, enhancing the architect's experience, and leading to more innovative solutions. Besides, they allow less tedious labour from the designer, leaving the experts with more time to dedicate to other, less quantifiable, tasks, such as the evaluation of designs' aesthetics.
Acknowledgment:
Gramazio Kohler Research, ETH Zurich:
Aleksandra Anna Apolinarska, Matthias Kohler
Swiss Data Science Center:
Luis Salamanca, Fernando Perez-Cruz
Project: Semiramis, ETH Zurich, Tech Cluster Zug, 2021-22