TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Judgment · AGE 18—21 · FACTUAL · DURABLE

Prompting is thinking about thinking

A vague prompt is usually a sign of a vague intent.

The idea

Working with these systems is mentally expensive. You have to know what you actually want, judge what comes back against it, and decide how far to trust the result. Much of that is not about the tool. It is about you watching your own thinking, which psychologists call metacognition. Some of it is about the tool — limited control, hidden instructions, randomness — and the bake-off is where the tool takes its share. But a vague prompt is usually a sign of a vague intent. When the output disappoints, ask "what did I actually want?" before "how do I phrase this?"

Why it matters

Students add words to a failing prompt the way they add lines to a failing drawing. Naming the intent first saves both the words and the cycles.

See it in the studio

A student types "make it more architectural" for the fourth time. The tutor asks what architectural means here. Heavier? More ordered? More like a building and less like a film set? The student pauses and says "I want the structure to show." That sentence is the prompt. It was also the intent, and it took four cycles to find because nobody asked for it first.

Watch for this

Mistaking fluency for intent. A long, detailed prompt can still be vague about what the work is for. Length is not the same as clarity. One sentence that states the decision you are trying to make beats a paragraph of adjectives.

Try it

Take one failed prompt from this week. Before rewriting it, write in plain words what you actually wanted: what decision the output was supposed to help you make. Then rewrite the prompt from that sentence. Compare the two prompts. The second is usually shorter.

Prove it

Explain why a vague prompt is often a sign of a vague intent, not a weak tool, with one example where stating the intent first changed what you asked for.

How it works

Metacognition, meaning knowledge of your own thinking and the monitoring of it, was described by Flavell in 1979 and has been studied in learning ever since. A 2024 paper from Microsoft Research applied the idea to generative systems directly. It argues that the difficulty people have with these tools is largely metacognitive (knowing what you want, evaluating outputs, and calibrating confidence), and that interfaces could be designed to support that rather than leave it to the user. The Lab's reading is simpler. The prompt is a brief, and a brief is a statement of intent. Work on the intent and the prompt writes itself.

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