Gather what your students actually care about

1

Research

Why this matters, and what good looks like

Placeholder content, pending authoring in a later content batch.

Interest is not the same as engagement, and neither is the same as a project worth three weeks. What this method produces is evidence about a particular group, not a topic.

Where the evidence comes from

A Framework for High Quality Project Based Learning

Placeholder pending authoring. Student voice in project design is well-supported practice; the AI contribution here is summarising, which is low-risk but also unvalidated.

2

Workflow

Brief it, steer it, check it

Bring what they said, not what you remember

Paste the raw responses. Summarising before you start throws away the disagreements, and the disagreements are the useful part.

Prompt 1 · Group

CONTEXT
I teach [subject] to [year group].
Here are raw responses from my class: [paste them, no names].

REQUEST
Group these into themes. Name each theme in the students' own language.
Then name the tensions: where do responses genuinely disagree?

QUALITY BAR
Do not invent a theme that fewer than three responses support.
Say explicitly which themes are thinly evidenced.

FORMAT
Each theme as: the name, how many responses support it, and one
representative quote. Then a short list of tensions.

Check before you use it

Did you remove every student name before pasting?
Is each theme supported by more than one voice?
Do the themes sound like your students, or like a curriculum document?

Stop rule

Do not build a project on a theme only one confident student named. If the evidence is thin, go back and ask the class again.

3

Reuse it

So the next one takes minutes

Suitable

Build a skill

Save this as a reusable instruction you can run on any project.

Not suitable

Build an agent

Chain the steps, with a checkpoint where you approve.

A single summarising pass does not justify a chained workflow.