Post-Human Aesthetics in Architecture: In Conversation with Matias del Campo

Post-Human Aesthetics in Architecture: In Conversation with Matias del Campo

Technology is disrupting the creative industry and it’s only getting better, and faster. Innovation in the architecture industry has never been as rampant as it is at this moment. The advent of artificial intelligence (AI) in architecture – the first genuine 21st-century design method – is changing the way buildings are imagined and designed. AI image generators like Midjourney and DALL-E provide an efficient and explorative way of conceiving architectural concepts. Generated in less than 5 minutes, these images unveil an interesting design aesthetic that is emerging. In an exclusive interview with ArchDaily, architect and educator Matias del Campo hypothesizes what the future of architectural aesthetics would be.

+ 2

In comparison to most professional fields, the construction industry notoriously lags behind in the adoption of technology. What happened last summer surprised del Campo – the explosion of AI image generators into the architecture discipline. The tool instantly seemed applicable and attractive to many architects, resulting in a social-media frenzy of mindblowing renders. “There is certainly a novel aesthetic evolving, what we can call a post-human aesthetic”, the architect proclaims.

Matias del Campo is a registered architect, designer, and educator. He is the co-founder of the architecture practice SPAN, Associate Professor at Taubman College of Architecture and Urban Planning, director of the AR2IL – The Architecture and Artificial Intelligence Laboratory at UoM, and an affiliate faculty member of Michigan Robotics, Computer Science, and Data Science. His research focuses on the application of artificial intelligence to architectural design and the culture that is emerging from this intersection. In conversation with ArchDaily, del Campo shares his insights on digital architecture, human culture, and design agency.

What do you mean by the term post-human? What is post-human design?

People tend to get confused by the term post-human, assuming that it means ‘after humans’, but that is far from the meaning. Post-human refers to an era we are entering where the sole agency of humans is in question. There are other players that help humans be creative and expand their abilities, artificial intelligence being an example. This era will see a shift in the way we live and affects every creative field like painting, music, and architecture. The idea of learning from human behavior and expanding one’s abilities as a human allows us to be a part of the post-human ecology. 

The post-human aesthetic seen in the viral Midjourney or DALL-E imagery does not come out of singular human will, but from a communal dataset of over five billion images. It involves collaborative inputs from both humans and artificial intelligence. This technology allows us to interrogate a multitude of layers of human culture from today up to the earliest image to exist. We have the entire history of architecture at our fingertips and a learning system to help us explore it.  What fascinates me is the latent space that exists in between this enormous set of data points. Just like we see the moon in greater detail with a telescope, AI uncovers new images that come from this latent space.

It doesn’t mean that we have created something new, only something we have not seen yet.

If the generated images are derived from existing data, is there scope for originality in design?

The idea of being ‘new’ is a philosophical question – is there such a thing as originality? There are different ways to observe what is new. If I buy a pair of shoes, they are new. If I invent a new architectural language, the answer is more complicated. With AI, there is always a question about imitation, copying, replication, and inspiration. AI relies on existing data, and to create something uniquely new out of it is very questionable. To create something original, neural networks – a method in AI that teaches computers to process data like humans – would have to extrapolate from the data, which they are not good at. Neural networks are great at interpolating between data to mimic information and create something similar. 

We, humans, have the imagination to put together a prompt for image generators that seem controversial or even inconceivable. What the AI does is generate images that are different, but not necessarily new. What comes out of these image generators may not be unique in the events of history, but there is enough novelty to provoke architects to push ahead with a new idea. That is what I think is great with these tools, they are machines that can expand our imagination and mind.

If there is no ‘new’ idea with AI, what does that imply about authorship? Where does this put the architect and their work?

I’m quite suspicious about the idea of authorship in general, it is a concept that was invented in the 18th century when it was very clear who the creator is. In our contemporary use of AI, it is not so clear anymore. When looking at AI-generated images, one wonders who the author is – is it the artist who came up with the idea to use the prompt, or is it the programmer who developed the algorithm? Is it the artists in the data set whose work was used to create the image? Going forward, I think we should instead imagine a world without authorship. We seem to care so much about authorship because it acts as a stamp for humanly produced content.

Architects seem obsessed with the idea of the “sole genius” who sketches a building on the back of a napkin. We need to collectively acknowledge that those times are over.

AI image generators come up with the final image first, rather than design a building from the inside out. Would the widespread use of this tool alter traditional design methods?

At this point in time, AI image generators are creating sketches for possible projects. We’re turning around the design process entirely, starting with generating fantastic renderings that clients are moved by, only to be challenged with executing the project to match the image. At the moment, it seems like a top-down design process, but it is actually a bottom-up approach since the generated image is not fully designed. We could use the images to quickly visualize ideas and seek inspiration while on the drawing board. Soon, it will be possible to train a neural network to identify architectural features in an image, like windows or spaces. This could allow us to generate detailed plans and drawings from an existing render. With architectural history at our fingertips, we can optimize a design by interrogating the data to provide better spatial solutions.

What will architectural aesthetics be dictated by in the future? 

What we are seeing a lot with the influx of AI-generated architecture renders is people’s desire to create something complex without having to model it themselves. The images seem recognizable at a first glimpse, similar in form and aesthetics to the buildings we see every day. There are some elements, however, that exhibit enough strangeness to provoke us to look at the image again. These machines are great at recognizing and putting together features in a remarkable way.

When I talk about post-human design, I mention the terms ‘estrangement’ and ‘defamiliarisation’ a lot. I think this is where we are going with aesthetics. Post-human aesthetics cater to our human understanding of what we think architecture is, while concurrently transforming it to create an idea that seems new. There is a whole culture emerging from the use of this novel toolset. I’m interested in uncovering an early 21st-century aesthetic that is informed by history, without imitating it.

Editor’s Note: This article was originally published on October 03, 2022.