TALAMANA · THE AI LITERACY MAP FOR ARCHITECTURE AND DESIGN · Foundations · AGE 14—16 · POSITIONAL · HELD

"The AI decided" — did it?

A model predicts, matches and produces, and those verbs keep the machine the right size.

Our position

"The AI knows." "It thinks." "It decided." Each of these sentences puts a mind inside the machine. The Lab's rule is plain: do not use a mind-word when a mechanism-word will do. A model predicts. It matches patterns. It produces an output. Those verbs are true of every system on this map, and they keep the machine the right size. This is a discipline of language, held on purpose. It is not a finding about what machines can or cannot have inside.

Why we hold it

The words you use decide the questions you ask. "Why did it decide that?" has no answer you can check. "What pattern produced that output?" has one. And a student who says "it decided" has already handed over authorship before the jury asks a single question.

The strongest objection

Whether words like understanding, representation, reasoning and knowledge can properly be applied to computational systems is contested across AI research, cognitive science and philosophy, and it is not settled in the Lab's favour. Some researchers use "knows" and "represents" as technical terms for things they can measure inside a model. Forbidding the words may make students miss real questions about what these systems do inside — and it may be the Lab's preference dressed as a fact.

What would make us revise it

If the field arrives at measured, agreed meanings for these words as applied to models — so that "the model represents X" becomes a claim a student can check — the Lab will allow the technical senses and keep the rule only for the loose, everyday ones. We watch the interpretability literature for that every edition.

See it in the studio

A jury asks why your massing looks the way it does. "The AI chose it" hands your authorship to a pattern-matcher. "I generated options and selected this one, for these reasons" is an answer an architect can give.

Watch for this

Your own shortcuts. "It understood my brief" — it matched your brief. "It got confused" — it produced an unlikely output. The shortcuts feel natural because the machine talks like a person. That is exactly why they are a trap.

Try it

For one day, every time you or a friend says "it knows / thinks / decided / wants" about a machine, rewrite the sentence with predicts / matches / produces / is set to. Keep the list. Notice which rewrites change what you would do next.

Prove it

Take three sentences that give the machine a mind and rewrite each so it says what the machine actually did.

How it works

Models are trained to continue patterns: words after words, pixels toward a description. Products are then tuned to sound helpful, which makes mind-language feel natural. Researchers call the habit anthropomorphism. It is old and human, and these tools are built in a way that invites it. UNESCO's AI competency framework for students puts a "human-centred mindset" first for the same reason: the person acts, the system produces.

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