Original image submission from participating artists Ruy Klein, of a neural network generated hybrid of gothic architecture and Le Courbusier’s La Tourette.
The notion of the beast is quite scary, unknown, and often brutish. _Architectural Bestia_, a new exhibition put on by the Southern California Institute of Architecture (SCI-Arc), exposes the idea of a beast within the concepts of mutation and machine learning. Curated by Hernán Díaz Alonso and designed by M. Casey Rehm, the exhibition features work by Atelier Manferdini, BairBalliet, Baumgartner+ Uriu Architects, Current Interests, Natou Fall, Florencia Pita & Co., Ramiro Diaz-Granados, Griffin Enright Architects, Soomeen Hahm Design, HDA-X, Kordae Henry, Kinch, Lifeforms.io, P-A-T-T-E-R-N-S, Ruy Klein, servo LA-Stockholm, SU11 Architecture + Design, Testa & Weiser, Tom Wiscombe Architecture, William Virgil, and Liam Young.
Through artificial intelligence, the work showcased in the virtual exhibition will be exposed to an ongoing state of transformation and mutation, in utilizing key sets of practices derived from not only architecture, but also art and fashion. The result prompts a discourse regarding a restructuring in the cultural systems of design, architecture, and art, largely violating an obsolete order.
_Flaunt_ had the opportunity to talk to curator Hernán Díaz Alonso and designer M. Casey Rehm about the process of creating the exhibit and their hopes for the evolution of architecture. _Architectural Bestia_ will run virtually through August 31, 2021. View the exhibition [here](https://archbestia.com/).
Original artist submission from Tom Wiscombe Architecture, of a physical model for the Korean National Writing Museum competition
**Talk about the inspiration behind the exhibit.**
**Hernán**: It started three years ago. We wanted to do a speculation around the role of collective individuality and the role of technology. There’s a pragmatic aspect of how to make an exhibition. It was important to include a lot of architects in it. The first iteration was very simple and was based on two frames, a fixed image, and in the bottom there would be a monitor connected to the computer by the whole series of artificial intelligence code running machines that Casey developed. The idea was about how the individual and the collective participate in designing. It started much more as speculation, in what this technology is doing and what it can produce. But there was also a playful attitude that we could have this artifact that could keep producing iterations. The third installation will be the finale, and we keep adding designers. The thing is evolving. It was to recognize how the students were working. It was intriguing to see students looking for traditions and other things. There are multiple designers all over the world working on similar ideas, so it was seeing if the machine could recognize that. It was a simple idea that became complicated over time.
**Casey**: We were kind of thinking about the way some of the machines were working and producing this exhibition where we’re using this neural network and process of automation as a lens of looking at the aesthetics that are supported by the institution that kind of curated artists from this. But then how that kind of juxtaposes against the aesthetics, which are kind of privileged on social media by maybe collective evaluation or like, ideas about image production outside of the Academy. And what's really interesting is I think the way we're using the machine. They're designed to make kind of really photorealistic images of deep fakes of faces, and they kind of need to be trained on a very narrow data set of images with like 20,000 images of faces, and then it can make a really good face. Instead, what we’re trying to do is see if we can use the machine to kind of assemble aesthetics or assemble features.
By training not just within the existing artists, it's not really their kind of final works that we curated into this data set. But we allowed each artist to define how they wanted the machine to learn about their work. So we have a really wide range of content that's going into it. So it's not just final building photographs or renderings. But there's a lot of process models and sketches and painting techniques, and giving the people participating in the show to really frame what it is they felt was critical in terms of representing their thought process and the design process to the machine. So it's been really interesting to see how the network's able to try to move around all of those seats as I try to assemble content from those machines. It's a very different lens than the way we used to look at the work it gets once it's bound by a very specific set of linear algebra. So it's going to have its own biases, and so an alternative perspective. But it becomes this project that allows us to look at huge vast spaces of design potentials, and just kind of continuously evolve and explore and wander around in that space.
Original image submission from participating artist, Elena Manferdini, showing a study of a synthetic garden
**Have you ever worked with machine learning before this exhibition?**
**Casey**: Yeah, it's a pretty central part of our office for the last 10-12 years. With neural networks, it was really not until about 2014, we started about six years ago with those.
**What was the process like of deciding which artists would be included?**
**Hernán**: It was a little bit of internal desires, and outside deadlines and boundaries in the original lineup. I think it was limited to 12 or 13 outdoors, because that's the space that they gave us. So in the first version of the show, the idea was to immerse people from SCI-Arc and the word of the faculty, that were obviously invested in technology and the use of technology in their practices and in their teaching. Everybody uses technology today, but they were the ones that they really were at the forefront, conceptually, theoretically, and so on. In the second round, as we learned with the process, and Casey kept developing more and more power into the thing, we thought it was interesting to invite a group of designers and architects that, at first glance, didn’t fit that mold. They work with different logics, different trajectories, different appetites in relation to technology. And it was interesting, because they start to do certain disruptions. In the third iteration, we add more refiners mixing those, but also we start to add people who, even though their formation is architecture, they are operating in other fields. So we invite people that are much more like filmmakers or makeup artists. We invited people who are much more removed from the core of architecture. The idea for this exhibition to keep evolving.
**Talk about the name of the exhibition, _Architectural Bestia,_ and how you came up with it.**
**Hernán**: It was because it was the notion of creating a jungle, how architecture is becoming this beast. Bad science fiction movies have these beasts that keep observing and mutating and evolving. It's more about the notion of any species than any notion of typology of pre-established order. There was a moment that we were talking about architectural django. But at the end of the day, it was manipulated through objects. So it seemed like Bestia was a more proper thing. But again, the beast is also mythological, some people are afraid of it, you can domesticate it, it can become your friend. So there was all this conversation. But also, we thought it was a fun title. Sometimes we used to say “This machine is a beast.” And I remember there was an echo of that too.
These ten images represent the site’s (archbestia.com) first 10 days of evolution of LA Bestia. Initially trained on 23 curated artists in the exhibition, the software continuously retrains the image generation GAN by incrementally adding images sourced for each day’s top social media posts on architecture. The center image of the grid is the machine’s attempt to reproduce the previous day’s most viewed image. A caption generating neural network was trained to produce texts for each image when moused over and enlarged.
**You said that you came up with the idea about three years ago. Do you find the element of mutation now being more relevant than ever, in the turbulent times we're living in?**
**Hernán**: I personally like the notion of mutation because mutation implies a simulation, implies intelligence, implies evolution. In the world of design, I am a little bit skeptical sometimes of the blind faith in technology. I've been interested in this for more than 25 years, and technology keeps changing. So the promise of technology becomes obsolete very fast. That's why I like mutation more, because I think mutation is something that keeps evolving and absorbing technology. Is that enough? Probably not. I think we have to start thinking of mutation in relation to what, and I think we have to be more specific about it. But I think, at least from my point of view, on occasion, we have a different view about that. And maybe the people included in the exhibition have a different view about it. Mutation, to me, is a very dynamic way to think about it and to not be attached to any particular technological apparatus, but understand it as a cultural thing, to think of it as a sponge, that keeps absorbing and changing with the times. Technology issocial, political, economical, and everything else. So yeah, to me, it still makes sense.
**Casey**: As we were putting the second round together, we really opened up the types of artists that were coming into the beast training network. And I think we're at a moment where we have a kind of overwhelming quantity of information and complex information, right? Like, everything's kind of operating at a global level. So taking a machine like this, which in some ways is kind of nihilistic at its outset, right? It doesn't have any of the preferences of an architectural history degree or theoretical discourse, it's this kind of equal opportunity consumer. I think a device like this, which is able to find adjacencies and relationships between extremely desperate works, or extremely desperate, aesthetic productions, in the full built out version, starts to show us relationships and adjacencies on a potentially global level as these types of devices or these types of software become more and more utilized. That can help us reframe the way that we're looking at a really complex existence and cultural phenomenon we're living through right now. And when we looked at how these tools are used, in general, they're often used to kind of optimize or govern or do things like automate sentencing or deal with contracts and things like that. So to begin to kind of leverage them for looking at how we can use them for generating cultural discourse or cultural production, I think is really critical, as well. I think it's an incredibly relevant project for the contemporary global society that we live in right now.
These ten images represent the site’s (archbestia.com) first 10 days of evolution of LA Bestia. Initially trained on 23 curated artists in the exhibition, the software continuously retrains the image generation GAN by incrementally adding images sourced for each day’s top social media posts on architecture. The center image of the grid is the machine’s attempt to reproduce the previous day’s most viewed image. A caption generating neural network was trained to produce texts for each image when moused over and enlarged.
**Do you think that the aesthetics valued by the architectural community differ greatly from the social media community? Do you notice very stark differences?**
**Casey**: Individual social media platforms will start to create different trajectories of aesthetics. You know, based on how people are pursuing likes and uploads. A theorist at NYU has a couple great books, where he did some deep dives on Instagram and did some image analysis and was able to kind of really dive into some of the trajectories that just emerge as people start to appropriate each other's kind of styles or the aesthetics of those styles that get more likes. I can make banal broad sweeping generalizations like Flickr privileges the SLR set, the big digital cameras, big lens, and you’re seeing a lot of castles and landscapes on the top coated architectural content. So in the third round, we're going to keep running it back by designing how the machine searches the internet, to try to keep it so it doesn't go down one of these specific rabbit holes of one specific platform. Yeah, so looking at how the machine searches the internet is a big part of the project. But definitely, there is a gap between what is privileged on social media versus what we privilege at the Academy.
**What kind of message or theme do you want the average spectator to gain from looking at the exhibition?**
**Hernán**: The expectation is always to engage a certain level and to really develop some kind of a conversation, like a blur feel. To me, an exhibition like this is always an opportunity to establish different kinds of dialogue of what architecture could be. So it's a strange thing, because it may not be recognizable through the eyes of what we think about architecture. But it's recognizable to other things that what we think are the images of things, or images of architecture. So at the end of the day, I think those aspirations are the same instance, to keep pushing the cultural relevance of architecture.
**Casey**: It becomes, I think, a lot less of us saying specifically if you combine these two designers, you'll get some kind of interesting aesthetic out of it and more about this kind of unbelievably large kind of endless Space of potential aesthetic production from culture that is much more open to interpretation. It's much more about emphasizing that kind of vastness potential of automation and less about us explicitly telling you how you should feel from it—hopefully, expanding what you think an architectural image should or could be.
Original image submission form participating artists, BairBalliet of a digitally fabricated wall mirror.
**On the notion of reinventing and growing from old ideas and mutating, do you think architecture is a cultural practice that needs to mutate in the future and over time?**
**Hernán**: Architecture is a slow conjecture that requires a lot of time and is not just slow, because it takes a lot of time to build things, but is slow because it's a discipline that feels very comfortable with being slow. I think many people consider fashion. It shapes society and culture faster than any other thing. Of course, because there's a basis, saying that you need to put out four collections a year, but nevertheless, they're never worried about the relevance. They're never worried about the importance. There’s an idea that buildings need to be timeless and all the things. So my hope is that yes, that is the case. There is very little evidence to believe that it will happen fast. I always joke that architecture is the elephant in the jungle, it’s slow and big. The cool thing is that it has certain intelligence and memory. And it's so big that they make it very difficult for other animals to attack.
I think there is a big dichotomy between the discipline and the profession. So whatever we call discipline, what I like to call the speculation versus the production of architecture, I don't think is going to be solved anytime soon. But I think an exhibition like this is one to show the potential of that. But culturally, I think we need to be much more dynamic than we are. But I think we are on the way so we can believe in optimism and also, again, there’s the whole issue of social issues and the racial issues and so on, which are not easy to tackle in architecture, but they need to be addressed. But it shouldn't be a separation, it shouldn't be a speculative design project, or a social project, like they are in two different camps. It should be intertwined. It's all part of the same thing. You cannot have one without the other one, the idea that aesthetics or design doesn't have a social value. So I don't think we have the answers. But I think they have to be committed to exploring that.
**Do you guys both hope to work with exhibitions related to technology and machine learning in the future? Is this an idea you want to explore further?**
**Casey**: It has become a big part of our practice. I've got another one coming up in Seoul. This fall, we're doing a piece exploring the way that the building is a kind of a physical part of a larger stack of technologies now, that are kind of governing and modifying and encouraging certain ways that we operate or occupy the city or the building. So we're looking a bit, specifically in Seoul, trying to speculate on how Korea is being really aggressive with their smart technology, smart city tech, surveillance, things that we would probably call surveillance here in the States, and the fact that they were able to manage the COVID-19 pandemic better than just about anybody. But I think there's a lot of questions about how these sites of technology start to change the way we live. I think there's something like the beast. How can we add complexity or sophistication to the tools for the design and production of architecture, but maybe underexplored is how these technologies are now becoming an embedded and ubiquitous part of the architecture that we design. This has become like a pretty big focus for us moving forward. We occupy the city completely differently now. Platforms allow us to bring things to us. Public spaces have taken on different meetings. So I think there's a lot to really explore about how technology is reshaping the built environment and the way we engage with it.
.jpg)
