TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Judgment · AGE 17—21 · POSITIONAL · HELD
Information is not experience
The felt, cultural meaning of a place is the layer you are there to carry.
Our position
A model can tell you a stepwell's depth, its date, its catchment and its function. It cannot tell you what it is like to go down into one in May: the temperature dropping step by step, the sound changing, what the place means to the town. AI processes information. It does not live experience. The felt, cultural meaning of a place is the layer you are there to carry, and no amount of information replaces having stood in it.
Why we hold it
"No site, no client" is the closed-world rule elsewhere on this map: the model has never stood anywhere. This card extends it one step. Even with every fact about a place, the model has no sense of what it means. In architecture, meaning is not decoration. It is the brief. The courtyard that is the centre of a family's life and the courtyard that is a leftover between blocks have the same dimensions. The difference is the thing a designer is hired to know.
The strongest objection
Serious people disagree about whether this is a permanent boundary or a current one. Models now take in images, video and sound as well as text, and some researchers argue that systems trained on enough of the world's recordings will come to hold something like experience, or at least a working model of the physical and social world that is not "mere information." On that view, "AI does not live experience" is true of today's tools and unproven as a principle, and a card that states it as a principle may age badly.
What would make us revise it
If a model could tell a designer something about the meaning of a specific place (not its average, not the mood of its photographs) that the designer then confirmed by going there and that no text had recorded, the line between information and experience would need redrawing. We would also soften the card if the philosophical disagreement moved towards consensus. It is marked as a held position and reviewed every edition because the disagreement is live.
Try it
Take a place you know well. List what a model can state about it: dimensions, dates, materials, function. Then list what only lived experience holds: what it sounds like at dusk, who uses it and when, what it means to the street. Explain why "no site, no client" extends to "no lived meaning."
Take it to crit
Can the student name the experienced meaning of a place that the model's information leaves out? Ask them about a place they have stood in. Then ask what of that made it into the brief.
How it works
The card is marked as contested inside the Lab's own position because the disagreement is real. Today's models hold descriptions: text, images, recordings. They do not hold a body that was cold on the seventh step, and their account of what a place "means" is assembled from what people have written about places like it. That is information about experience. It is useful, and it is not the same thing. Whether future systems (embodied, continuous, learning from the world rather than from records of it) close the gap is an open question. This card names it rather than settles it.
What this idea builds on
What this idea opens up
- Nothing yet names this as a foundation.
Sources
Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.