GHASSAN ALSERAYHI
 MSc. Arch, M. Arch, B. Arch, Assoc. SCE

ARCHITECT + EDUCATOR + RESEARCHER
ALIENATED MOQARNAS MACHINE LEARNING FOR HISTORICAL DATA-INFORMED GENERATIVE DESIGN

SUBJECT: ‘AI and ARCHITECTURE’
TYPE: DESIGN RESEARCH SEMINAR
INSTRUCTOR: MATIAS DEL CAMPO
DATE: 2021
The seminar aims to familiarize students with a range of ideas and methodologies for employing Artificial Intelligence in architectural design. With AI becoming increasingly influential in architectural design processes, the seminar opens up a variety of investigative avenues. It covers the application of Machine Learning for structural optimization and delves into the ethical aspects of using AI in this field. The course is bifurcated into two main sections: practical skill development and theoretical understanding. 

In the practical segments, students are introduced to various AI tools relevant to architectural design, followed by hands-on exercises involving user-friendly applications, cloud-based platforms like GitHub, Google Colab, and the utilization of online data sets. The theoretical part of the seminar addresses the broader implications of AI in architecture, discussing issues related to creative ownership, agency, the feeling of estrangement, and the biases inherent in data. Lectures in this segment also explore the use of AI tools in related disciplines such as the visual arts, music, and literature.

In this seminar, the work criti­cally investigated the role of Al in a future world of building, in­cluding the conversations on the impact as a cultural tech­nique for the production of architecture. Also, it interrogated the consequences of the implementation of learned Al tech­niques used in architecture pertaining to its role as an agent of culture. Questions have been discussed such as “Are we start­ing to share the agency when it is about design decisions?” “How does such a design conglomerate contribute to architecture as an agent of culture?” “Is this the emergence of a new understanding of the production of culture. Is there something like Al culture?”







Examples of the original pictures used in the process:                                                                                                                                                                                        
                                       


                                       





Example of the results after scripting:

                                       


                                       





Using StyleGAN Transfer 2D to Generate Liminal Space Walkthrough

Over 1000 pictures have been collected to the database that are used in the script­ing of the StyleGAN transfer 2D. The task given required us to pick an architectural element and experiment with it using artificial intelligence, basically machine learning to explore the limita­tion of new features, geometries, possibilities might come out of the process. Then transferring the results to entities that might carry/delever spatial qualities. In this assignement, Islamic Muqarnases from different times and places were picked randomly as a raw data to be then, experimented with StyleGAN Transfer technique.
































Using Pexils to Convert Images From 2D TO 3D

In order to translate the previous results to deleverable language that carries architectural features, Grasshopper scripting used in rhino after using one of the final pictures as a material, using bitmap and material material displacement features in Rhino.