Reading
AI images are a mirror that shows human desire without hesitation.
AIの画像は、人間の欲望をためらわずに映す鏡だ。
One AI image, a few numbered points. Tap each one to see what an architect who has built notices: what would stand, what would fail, and what the AI forgot.
1234
Reading 01 · 4 points
The AI put a waterfall inside a house on a cliff. Here is what an architect who has built on cliffs sees in it.
1234
Reading 02 · 4 points
Under the sea the AI draws architecture that looks like marine life. Read closely, it is also a pressure vessel.
123
Reading 03 · 3 points
It took thirty seconds to draw. In 1990 a perspective like this took me a week. What would it take to build?
1234
Reading 04 · 4 points
The AI drops an organic tower between ordinary Tokyo buildings. The rules that shape every real Tokyo street are missing.
1234
Reading 05 · 4 points
Coloured glass boxes stacked on a street corner. Beautiful from the crossing; harder to live in.
1234
Reading 06 · 4 points
Two arched towers rise from the water with a city behind them. Structurally they are closer to bridges than to buildings.
1234
Reading 07 · 4 points
When I first saw Blade Runner, it was the future. Now it is here. Yet the megacity the AI draws looks less like a city no one has seen than a mass of ideas we already know, like an old science-fiction film.
1234
Reading 08 · 4 points
When the AI made a house with the genes of De Stijl, the hard edge of that era was gone. I saw instead the delicacy of a Japanese car. The colours and mood were inherited, but a completely different house was born.
1234
Reading 09 · 4 points
When I was a university student, Zaha Hadid was called a paper architect, her designs said to be unbuildable. Even to a student's eye, her work was avant-garde, unlike anything by Japanese architects. She went on to build, and many have inherited her genes. Type an avant-garde prompt into the AI, and even without entering her name, it inherits Zaha on its own.
1234
Reading 10 · 4 points
In 1990 I joined the office of the architect Itsuko Hasegawa. She built public buildings that brought landscape and architectural art together, what she called a "second nature", and went on to make many buildings with sky gardens. Having worked on one of them, the Niigata City Performing Arts Center, is one of the great treasures of my life. The rooftop garden is still a mode today, spreading into many buildings, and the AI has learned this well.
1234
Reading 11 · 4 points
Architecture that wears a soft curtain as its outer skin brings to mind a woman with thin skin. Unlike the hard, heavy, powerful architecture of the classical tradition, this is fragile, beautiful, feminine architecture. It is like mizu-manju: sweet bean paste wrapped in translucent jelly, a sweet that is cool and delicious.
1234
Reading 12 · 4 points
I grew up watching Leiji Matsumoto's Galaxy Express 999, longing for spaceships with a mechanical look. Not a Tesla but a DeLorean; in Star Wars, not Padmé Amidala's smooth Naboo starship but Han Solo's machine-like Millennium Falcon. The boy became a student and, by luck, landed a summer open-desk at Makoto Sei Watanabe's office, making models for the Aoyama Technical College. It lasted only a month, but turning sketches into plastic models was thrilling. Those mechanical bumps and ridges take a frightening amount of time; for anyone who loves building plastic models, that time is irresistible. AI is taking that thrilling time away from us. Architecture's value runs against being finished quickly. Like Gaudí's buildings or a fine spirit, people believe what takes longer is worth more, and AI, by the nature of its trade, cannot understand this. AI and Han Solo's Falcon are both asked for one thing: speed. It is ironic: cutting the time turns a work of art into something as worthless as a pebble.
1234
Reading 13 · 4 points
Humans could not keep to buildings that sat firmly on the earth and rose honestly toward the sky. Like Frank O. Gehry's Guggenheim Bilbao, architecture breaks away from its old structure. In other words, humans cannot keep doing the same thing, or what was once defined as right. If AI is learning from us, it will understand that humans want to break the world that was defined as right. That is exactly the fear: that AI might one day rise against us. Looking at humanity's long history of stability and instability, AI may judge that this is what we want. There is always the danger of breaking away from a stable structure.