TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · The Brief · AGE 16—18 · POSITIONAL · METHOD
Stated, inferred, unknown, assumed
Label every line, because an unlabelled brief turns its assumptions into facts within a week.
When to use
Every time a brief, a survey, a client email or a generated draft arrives and you are about to design from it. Especially when a model has written part of it, because a model writes its guesses in the same tone as its facts.
The method
Four labels, one per line of the document. Stated: the client or the source said it, and you can point to where. Inferred: nobody said it, but it follows from something that was said, and you can show the step. Unknown: it matters and nobody has said it. Assumed: it is unknown, you need a value to proceed, so you have chosen one for now and written the choice down. Then apply one rule. If an unknown would change the design, it is asked, not invented. Do not invent north. Do not invent occupancy. Do not invent the budget. Take those to the client as questions. Keep the four labels on the document, because a brief whose lines are unlabelled turns its assumptions into facts within a week.
Watch for this
An inferred line that has lost its step. "Family of four" became "three bedrooms" became "no study", and nobody can now say where the third bedroom came from. If you cannot show the step, the line is assumed, not inferred. Label it so.
The Lab's note
The four labels are the Lab's method, and the rule about north and occupancy is ours. It sat inside the generate-the-brief card and we have lifted it out, because it applies to every source, not only to a generated one. Other studios keep a single assumptions list, which is a good habit. We think the list must also say which lines were inferred, and from what, or the assumptions spread.
Try it
Take a generated brief, or a real client email. Write the four labels down the margin, one per line. Count each label. Then pick the three unknowns that would change the plan most and write them as questions to the client. Most first attempts find that the stated column is short, and that at least one assumed line was about to become a wall.
Prove it
Show one brief with every line labelled, and the list of unknowns you sent to the client instead of inventing them.
How it works
A model asked to complete a brief completes it with the most probable content: probable plot sizes, probable budgets, probable orientations. The four labels are a way of reading that output without being fooled by its tone. They also describe every human brief. Peña's problem seeking is, among other things, the discipline of finding out what is not yet known before you design, and a list of open questions is one of the things it produces. The red-pen habit on this map is the same method applied to one kind of document. This card is the general form.
What this idea builds on
- A starting idea.
What this idea opens up
Sources
Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.