Marking and feedback prompts

After school on Wednesday, a Secondary 2 maths teacher is marking books in the staff room. She pastes one student's working into a general AI and asks it to mark the script. Three seconds later she has 16/20, neat comments, and a fully correct solution she did not ask for. She is about to copy the comment onto the mark sheet when she stops: the student never wrote the step where the unknown is defined, and the scan still shows a student number in the corner.

Marking prompts should give AI the job of provisional feedback while the teacher keeps the mark and the final comment. This lesson follows lesson 8 and concentrates on writing rubrics, rewrite bans and review requirements into the prompt itself.

---

Provisional versus final

Marking is two jobs: judging where the student sits, and telling them what to do next. Models are usually strong at the second, because it is pattern description and gap-spotting. They depend entirely on you for the first — with no rubric, the model marks against an average impression of student work, which may bear no relation to your school's standard and may drift between scripts.

Think of it as spell-check: it can flag a suspect word, but it does not decide what the essay is worth. The teacher is still the gatekeeper. AI prepares a draft so you spend less time hunting for problems, not less time thinking.

Every marking prompt therefore needs three things written plainly: the rubric, a ban on rewriting the student's work, and a label that the output is provisional and requires teacher review. Our article AI marking and instant feedback already covers three subject scenarios and the checkpoints teachers must keep. This lesson does not repeat that; it concentrates on the prompts.

De-identify before you paste

Before writing any marking prompt, deal with student data:

• Strip or replace identifying data: names, class, student number, parent names, addresses and internal school IDs — use "Student A" or a number instead.
• Do not paste a spreadsheet with thirty names and marks into a general tool; that is a personal data list, not marking material.
• Work on a gated school platform: four-layer gating, teacher monitoring, encryption at rest and an ai_audit_logs trail, none of which a general chat tool has. Our overall approach is on the security page.
• Tell students: AI is the first reader; the teacher is the final marker.

For how safety settings get circumvented and what actually holds, see lesson 14.

Three subject templates

The structure is the same across subjects; square brackets are filled in by the teacher for each assignment.

Chinese expository writing against a rubric

You are a marking assistant for a Hong Kong primary Chinese Language teacher. Your output is a provisional assessment for teacher review, not a final grade.
Task: provide a provisional assessment and feedback on one student expository text against the rubric below.
Context: Primary 6 expository unit, learning focus is explaining a procedure in steps with connectives showing order. The comment will be read by the student, so use language they can understand.
Rubric: content 6 marks (complete steps, concrete detail); structure 5 marks (clear paragraphs, logical order); language 5 marks (accurate word choice, appropriate sentence patterns); punctuation and spelling 4 marks.
Output format: three parts — (1) a four-row table, one per dimension, with suggested mark, one sentence of justification and a quotation from the student's text; (2) the comment for the student, three to four sentences, the first naming a specific strength with a quotation, the second one area to improve, the third one action they can take today; (3) one line headed "for teacher review" listing the two judgements you are least certain about.
Constraints: do not rewrite any of the student's sentences, only indicate a direction; every quotation must be words the student actually wrote; all marks are provisional and the output must open with "provisional — requires teacher review before release"; do not reproduce personal data from the composition.

Student text (de-identified):
[paste]

English writing with marking codes

When an English Language teacher marks writing, the prompt should be in English and the marking codes must match the school's own.

You are a marking assistant for a Hong Kong secondary English Language teacher. Your output is a provisional assessment for teacher review, not a released grade.
Task: give feedback on one piece of student writing using our marking codes.
Context: Secondary 2, a 200-word letter to the editor on reducing plastic waste. Learning focus is clear opinion, supporting reasons and a closing call to action. Students are Chinese first language; common problems are article errors, tense shifts and comma splices.
Marking codes: T tense, SV subject-verb agreement, ART article, WW wrong word, CS comma splice, ^ missing word, ORG organisation.
Output format: three parts — (1) the student's text reproduced exactly as written, with codes inserted in square brackets at each error point and nothing else changed; (2) error counts by code, most frequent first, naming the two the student should fix first; (3) three sentences of feedback to the student, one strength with a quotation, one priority, one concrete next step.
Constraints: never rewrite the student's sentences or supply a corrected version — insert codes only; do not alter spelling or punctuation in part 1; any suggested marks are provisional and labelled as requiring teacher review; do not reproduce name, class or student number.

Student writing (de-identified):
[paste]

Diagnosing mathematical working

You are a marking assistant for a Hong Kong secondary Mathematics teacher. Your output is a provisional assessment for teacher review.
Task: diagnose a student's working, find the first step that goes wrong, and infer the likely misconception.
Context: Secondary 2, unit "applications of Pythagoras' theorem". The standard method has four steps: sketch and label the right-angled sides, set up the equation, calculate, state the answer with units.
Output format: three parts — (1) a three-column table comparing the student's steps with the standard method, marking each correct / incorrect / missing; (2) one sentence naming the first failing step and the misconception you infer; (3) two practice questions of the same type with different numbers, to test whether the misconception is cleared, with answers.
Constraints: do not rewrite the full solution for the student; do not assume steps the student did not write — mark them missing and note "teacher to confirm with student"; if the student's method is non-standard but mathematically valid, say so explicitly; all judgements are provisional and require teacher review.

Student work (de-identified):
[paste]

Feedback is valuable when it shows the student where they are stuck and what to do next — not when it hands them a better paragraph to copy.

Whole-class error patterns

This prompt often has more instructional value than individual comments, because it answers what you teach next lesson.

You are a data analysis assistant for a Hong Kong secondary [subject] teacher. Your output supports a teaching decision.
Task: summarise class performance on one assessment and identify shared error patterns.
Context: [year level], unit "[name]", class of [number]. Learning focus numbers and mark allocation: [paste specification]. The data I supply is per-item error type counts, with students identified by number rather than name.
Output format: four parts — (1) one row per learning focus, giving class mastery and the two most common errors; (2) three teaching recommendations, each naming which learning focus the next lesson should address first and why; (3) a suggested tiering, estimating how many students need part one, part two and the extension; (4) one sentence identifying which conclusions are limited by sample or item count and should not be over-read.
Constraints: reason only from the data I supplied; never name individual students; where a learning focus is covered by a single item, state that the conclusion is weakly supported; do not recommend anything requiring extra lesson time.

Data:
[paste]

For General Studies and Science practical reports, ask for feedback against the four stages of inquiry. For Visual Arts appreciation writing, ask the model first to identify which visual-element vocabulary the student used, then say whether the student is still describing or has moved into interpretation. For SEN adaptation, add an explicit constraint not to treat spelling error count as the main judgement and to report punctuation and spelling in a separate section.

Where people go wrong

• Asking for a mark without a rubric. Severity drifts between scripts and bears no relation to your school's standard. Self-check: does your prompt include a full rubric or marking code list? If it only says "mark this", there is no standard.
• Letting AI produce a rewritten version. Students copy it and the feedback becomes pointless. Self-check: does the constraints line forbid rewriting? If the output contains a full corrected paragraph, the prompt fails.
• Treating provisional output as final. Responsibility stays with the teacher. Self-check: have you sampled a portion fully yourself? Does the output open with "provisional — requires teacher review"?

Try it once

Take one student script you have already marked (single student, de-identified), choose the closest template above, swap in your school's actual rubric, and run it once.

What to do: Place the AI provisional assessment beside your own marking. Compare item by item: is the mark gap acceptable? Did the feedback rewrite anything? Did it quote the student's actual words?

What to hand in: Three lines — (1) how many marks the AI differed from yours; (2) which one comment was most useful; (3) which part you had to rewrite or delete.

How you know it worked: The provisional output saved you time finding problems, but nothing could be pasted to the student without your edit or confirmation. If AI rewrote a whole section, strengthen the constraints line.

After marking, a student may still need step-by-step guidance to improve on their own. That calls for a tutor that gives hints rather than answers — see lesson 10, scaffolded tutor prompts.