AI in Schools
Generate twin exemplars to teach the gap between A and C
Generate twin exemplars to teach the gap between A and C
Comparing a strong response to a borderline one, side by side on the same task, lifts student writing faster than another round of red pen. The bottleneck is that writing two student responses takes a teacher most of a free period. This prompt does it in two minutes.
The prompt
You are an experienced [SUBJECT] teacher and assessor in an Australian secondary school. You have moderated this task many times.
The task: [PASTE THE EXACT TASK AS YOU SET IT FOR STUDENTS]
Year level: [YEAR LEVEL]
Curriculum framing: [ACARA v9.0 achievement standard, VCE performance descriptor, or your own success criteria, copied verbatim]
Word or time limit: [e.g. 250 words, or 40 minutes]
Produce TWO complete student responses on this task.
Response A: an "A" standard response for this year level. Confident, correct, with the specific moves that distinguish strong work in this domain. Do not write like a teacher. Write like a strong student.
Response B: a "C" standard response on the SAME task. Plausibly written by a real student at this year level, not a parody. The errors should be representative of what genuinely shows up in marking, not random. Same length as Response A. Australian English in both.
Then produce a "What shifted" table. One row per specific move that took B to A. Columns: Move, Where you see it in Response A (short quote), What Response B did instead (short quote), Why this matters against the criteria above.
Finally, one short paragraph titled "Next teaching move if a student handed me Response B". Be concrete and small. Not "scaffold their thinking". Give me the actual sentence stem, question, or five-minute activity I would use in the next lesson.
What it does well
Forces the model to commit to named, specific moves (a topic sentence that takes a position, evidence that is interpreted rather than summarised, a counterclaim handled in one clause) instead of the generic "use more sophisticated vocabulary" feedback that helps no one. The "Next teaching move" line is the bit you can actually do on Wednesday morning.
Where it falls over
The "A" response will often read slightly too polished to pass as Year 9 work. Ask the model to "dial the writerly voice back by 20 percent and add one small imperfection a real student of this age would make". Also, the quality of the "What shifted" table is only as good as the criteria you paste in. Vague success criteria yields vague moves.
Try it like this
For Year 9 history (extended response on causes of an event): paste the SAC-style question and the VCAA-style performance descriptors verbatim. The "what shifted" table tends to surface the cause-versus-trigger distinction students miss.
For Year 4 persuasive writing: paste the NAPLAN-style prompt and the official NAPLAN writing marking guide for persuasive texts. Use Response B with the whole class as a shared edit.
If you only have ten minutes today, run it on the task you are setting next week, then ask the model for a third "B-plus" response that bridges A and B. That third response is your worked example for the modelling phase of the lesson.
Two exemplars and a table beat a paragraph of written feedback most weeks. Use the time you save to read the names of the kids in C standard and ask whether the model called it right.