TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · The Brief · AGE 17—20 · POSITIONAL · HELD

Name the drawing convention

A named convention packs a hundred decisions into one term, and an adjective packs none.

Our position

After the words for space and the words for material and light, there is a third vocabulary: the discipline's own drawing conventions. Poché on the cut wall. Beyond in fine-line elevation. Orthographic severity. Editorial presentation, not photoreal. Name these in a brief and a capable model can steer toward them, where the convention is represented in what it learned. Stay at the level of mood and it falls back on its default — extra trees, soft gradients, generic tastefulness. A named convention asks for the look. It does not make the model understand the rule behind the look.

Why we hold it

Poché is not black shading. Hand-drafted ink-over-watercolour is not watercolour illustration. A convention packs a hundred decisions into one term; an adjective packs none. When a brief drifts, the instinct is to add adjectives. The better move is to name the convention that already contains what the adjectives were reaching for. This is architecture's real advantage in prompting: the field has a vocabulary most other prompters do not. One line to keep beside it: a written instruction asks; the system's own controls enforce. The convention in the brief steers the look. What the section actually cuts is checked against the plan it was cut from, by you.

The strongest objection

Even where a convention is heavily represented, reproducing its look is not evidence that the model understands the rule that gives it meaning. A wall rendered black is not proof the model knows what the section plane cuts; a dimension string is not proof the dimensions agree. A model can mimic an architectural drawing and fail architecturally inside it, and the better the mimicry, the harder the failure is to see. The regional problem stands too: conventions thin in the training data return a costume. But the deeper objection is this — the model may know exactly what our convention looks like without knowing what it means.

What would make us revise it

Evidence that named conventions steer no better than well-chosen adjectives on the plates students actually make — or that the look they produce comes so often with wrong geometry that the steering is worth less than the check it demands. We already teach the check; if the check turns out to be the whole lesson, this card folds into the spatial check and stays only as vocabulary. Reviewed each edition.

Try it

Take a vague, mood-based prompt for a section or an elevation. Rewrite it naming three drawing conventions: the cut, the beyond, the presentation register. Run both. Then check the convention-led output against the plan: does the poché sit where the section plane actually cuts? Then try a convention from your own region's drawing tradition and find out whether the model knows it.

Take it to crit

When the brief drifted, did the student reach for the discipline's drawing conventions or for more adjectives — and did they check what the named convention drew against the plan it was cut from?

What this idea builds on

What this idea opens up

Sources

Open this idea on the map · The complete map · Logika · RBDS AI Lab, India · revised every edition.

Age grows from 11 at the centre to 22 at the edge, and six sectors show the learning strands. Tab into the map and the arrow keys step from idea to idea, following the links where there is one. Enter opens the idea under the cursor, and E reads out its links and the reason recorded on each. Press slash for Search, question mark for the full key list, and Escape to leave. Open Ideas for the complete readable list, including what each idea builds on and what it opens up.

LOGIKA · RBDS AI LAB INDIA
ON-RAMP · AGE 11 · FIRST ENCOUNTERS, NOT GATES — IDEAS · — DEPENDENCIES
DONE
OPENS NEXT
SOLID — STANDS ON