In this lecture, Matias del Campo unfolds the captivating realm of AI-generated images, as articulated in his book, "Diffusions in Architecture". The book serves as a compelling chronicle and speculative inquiry into the transformative impact of AI image generators on the architectural imagination, featuring contributions from six theorists and twenty architects.
Leveraging Ludwig Wittgenstein's axiom, "The limits of my language mean the limits of my world", the lecture delves into the contemporary surge of natural language text-to-image applications like Midjourney, Stable Diffusion, and Dall-E 2, revolutionizing the architectural landscape. This novel design tool has yielded a prolific output of extraordinary images and instigated theoretical explorations, hinting at the inception of a posthuman design methodology.
The confluence of these factors marks a seismic shift in architecture, redefining the dynamic between human and machine in the creative process. The lecture navigates the transformative impact of large language models, exemplified by ChatGPT, rendering promptism and prompt engineering obsolete. It has made it feasible for even the most inexperienced user to come up with functional code and complex image prompts.
All of this resulted in the proliferation of astonishing images, engendering a new epoch of design that blends human creativity with algorithmic intelligence. In doing so, diffusion models have emerged as a possible new design tool. Enabling architects to mine the multilayered, deep historical repositories of architectural knowledge for chimeras, capriccios, and mutants. Encouraging architects to discover a new voice for the architecture of the 21st century - one that is rooted in bold and visionary experimentation by default.