AI is not interesting to me because it replaces architectural judgement. It is interesting because it changes the tempo of visual thinking.
01 / Context
Before a design is precise, it is often atmospheric
At the beginning of a project, the question is often still soft: what kind of place is this, what should it feel like, what visual direction is it asking for?
This is where AI becomes useful. Not as the author of architecture, but as a tool for testing mood, light, framing, vegetation, material softness or the emotional temperature of an image.
AI does not decide what architecture should be. It gives the architect more material to think with.
02 / Tool
A fast sketchbook for visual direction
The first visual test used to be expensive: technically, mentally and in time. AI makes this early phase lighter. It can show several atmospheric options quickly and make them discussable.
An image as the start of a conversation
AI is strongest when it helps name something still vague. A project may want to be quieter, greener, more mineral, more open, less polished or more grounded in the terrain.
03 / Discipline
Without architectural control, it is only style
A good architectural image still needs scale, spatial logic and a relationship to the actual design. AI output becomes useful only when it is brought back to intent: site, section, massing, landscape and the story the image needs to tell.
Scale
The image has to keep a relationship to the body, the room and the real situation.
Intent
Atmosphere should serve the design, not cover it with effect.
Selection
Value comes from curatorial judgement, not from the number of variants.
Conclusion
Then the work has to become architectural again
AI can open a visual conversation. It can accelerate searching and offer unexpected possibilities. But after that, architecture has to return: precision, site, spatial logic and human judgement.
That combination is what interests me: not AI as a shortcut around work, but AI as a new space for thinking.