Despite the justified controversy surrounding AI art, you don’t have to worry about usurpation by software that can generate images of buildings, says Will Wiles.
These are uncertain times, but we can be sure of two things. The first is that art made with artificial intelligence (AI) is here to stay. (Many will feel that the word “art” in that sentence needs a protective embrace of quotation marks; some will feel that “artificial intelligence” does as well. Feel free to fantasize about those marks if you prefer.)
The second is that AI art will remain controversial, and rightly so. Human artists fear, quite reasonably, that they will consume much of the bread-and-butter work upon which they depend. To add insult to injury, artificial intelligence is something of a plagiarism.
However, she is here. Barriers to what to use are falling: In September, OpenAI ended the waitlist for its image creation platform Dall-E, so it can now be used immediately by anyone who signs up. And earlier that same month, an image Jason M Allen created using Midjourney, another artificial intelligence art application, won the digital art category at the Colorado State Fair’s fine art competition.
Art created by AI is here to stay
Sites like Midjourney, Dall-E, and Stable Diffusion use artificial intelligence to create original images from text prompts. The “intelligence” involved is more about learning, computational brute force, than intuition – the technique essentially relies on training computers to scramble images and solve them until they “know” how to create an image out of noise, as Daniel Finn writes in this helpful explainer.
Even at their best, the images created have a distinct clay quality, and some subjects, such as the human hand, seem to defeat it. Still, it’s good enough to spark widespread fears that it will permanently deprive underappreciated human artists of much of the business they depend on, while ripping off their work as human users tell software to imitate the work of creators they love.
AI engineering seems like a distant possibility. Architecture, after all, is a field that extends far beyond the visual—one day a single program might visualize a building, make it a set of constructible drawings, detail the drains and gutters, and work all the time with a client and a variety of contractors and administrators, as We expect from a human architect.
But is there really evidence that software-generated images of buildings are infringing on the architect’s remarkable ability to generate architectural ideas? Not that it may remain unfounded that concepts generated by artificial intelligence will be without effect – paper engineering can be a great moment.
Allen’s award-winning AI-generated portrait, Théâtre D’opéra Spatial, consists of costumed artists in a large, cavernous interior dominated by a massive circular opening. And more and more people are using these platforms to create architectural images along the same lines.
There are more than 32,000 photos on Instagram with the hashtag #Midjourneyarchitecture, and the Instagram account Midjourney Architecture, which collects interesting examples, has more than 34,000 followers. The “architecture” on display is mostly fictional, but sometimes eerily real.
An increasing number of people are using these platforms to create architectural images
Sometimes, Midjourney’s architecture can be compelling enough to create ripples in the real world. California-based Egyptian artist and designer Hassan Ragab used Mdjourny to experiment with ideas, but also to create an exaggerated homage to the traditional Islamic architecture of Cairo and Alexandria.
One of the images from the “Sketches of Cairo” series was viewed by 8.5 million Instagram users, a wall of windows in the shape of Escher, connected by intricate and implausible stonework and decorated with Islamic motifs. On Facebook, a post went viral claiming that this photo was from a real building, and journalists started calling Rajab asking him to check if it was real or not.
“It went beyond Egypt: people in Iran claim it is a Persian cemetery, and others claim it is a Spanish building,” Ragab wrote on Instagram. “Also, what was really ironic was how people were arguing (supported by their own beliefs and ideologies) about what this building had to do with.” News stories verifying the photo’s authenticity followed, including from AFP.
It usually takes no more than a second to determine whether or not an image is emerging from AI, but it seems few netizens can avoid that second, especially in the butchery of truth that is Facebook. So there is potential for confusion and intentional victimization – but not of a new kind, even if it uses new technology.
Right now, AI engineering is simply another form of cybernetic software: brilliant, never-realized concepts and dubious utopias that spread across the internet like visual fungus. But even those fungi have little nutritional value, as small, ambitious design practices and studios make use of them to promote their work and attract attention.
The same is true of fantasy architecture created as art by energetic and often brilliant digital artists who gather in places like the DeviantArt online community. If some people mistakenly take pictures as real buildings, or projects that may soon materialize – if some studio ingeniously encourages this impression in their version – it is a matter of their conscience.
Right now, AI engineering is simply another form of electronic software
AI engineering slots next to these areas, which leads to shakier results. Allen’s architectural background is superficially impressive, but there is less to it than meets the eye. It is just an impression of the baroque decoration and sophisticated structure, an interpretation lacking in the original. It has quirks that can be seen throughout AI engineering: distinct distortion; Inability to follow the line. My blindness is weird about symmetry. and misleading, impossible edges, connections and details.
And, of course, we remain anchored in the realm of the visible, not the real. The machine uses pixels in two dimensions, not materials and space. Architects can rest assured that AI has not yet come to their jobs, although it is not impossible to imagine AI portraits winning a student or a drawing competition, if that is permitted.
In general, architecture tests new technologies at the forefront and then construction uses them in the boiler room. Prefabrication briefly allowed metabolicists to dream of infinite reconfigurable flexibility, and today it offers cheap standardization. CATIA is published by Gehry and becomes part of the background. Promised by parametric abundance, 3D printing for use in service centers or something similarly unattractive will likely end up.
AI is more likely to find a role generating apartment floor plans or resolving service layouts—that is, exactly the areas where computers are already heavily used—than any visual creative application.
Insight is naturally what catches the eye, and it’s well worth looking into the cuteness of the photos on #Midjourneyarchitecture. That may start to view some of the above quirks as qualities rather than flaws.
Obviously, the results depend on human user input—which, in general, we don’t know—but interestingly there is frequent resonance with the work of a few particular architects with very strong individual styles: Gaudí’s biomorphism, the quirkiness of Eisenmann’s geometry, liquid iron.
Even more tempting, however, is that AI is alien to purists. The fascination it generates must be daunting to hardline architecture theorists and stylists, who focus on rigidly objective notions of beauty, and who regard deviation as an ethical as well as aesthetic failure. The device varies. Generative geometry is completely degraded, and it’s all the more interesting for it.
Will Wiles is a design writer and author of four novels Recently the last blade druid.
Top photo: Hassan Ragab.