GenAI can be a helpful assistant during assessment. It can save time on repetitive writing, improve consistency, and support clearer feedback. But it should not “grade instead of the teacher”. The safest and most useful approach is: GenAI drafts, the teacher decides.
Assessment is sensitive work. Students deserve fairness, transparency, and feedback they can learn from. GenAI can support this – if used with good habits: clear criteria, evidence-based comments, and careful human review.
1. Start with the rubric, not the assignment
GenAI performs better when the expectations are clear. Before looking at student work, it helps to have a rubric (even a simple one). GenAI can help you create or refine a rubric, but the criteria should reflect your course outcomes.
Good habit: ask GenAI to turn learning outcomes into a short rubric with levels (excellent / good / needs improvement).
Bad habit: ask GenAI to “grade this essay” with no criteria.
Example prompt (rubric draft):
“Create a simple rubric for a 500-word reflection essay. Criteria: understanding of concept, argument clarity, use of examples, structure, language. Use 3 levels and short descriptions.”
2. Use GenAI for feedback wording, not for the final grade
The biggest time sink is often writing similar feedback again and again. GenAI is excellent at drafting feedback text – especially if you provide the rubric and ask for specific, actionable comments.
A good workflow is:
- teacher checks the work and decides the main judgement
- GenAI helps draft feedback in a consistent and supportive tone
- teacher edits and finalises
Example prompt (feedback draft):
“Draft feedback for a student based on these notes: [paste your notes]. Tone: supportive and clear. Include: 2 strengths, 2 improvements, 1 next-step suggestion. Keep under 120 words.”
3. Ask for evidence: “Show where you see it”
GenAI can sound confident even when it is wrong. To reduce this risk, ask it to reference specific parts of the student text. This makes feedback more grounded and easier to verify.
Example prompt (evidence-based comments):
“Using the rubric below, identify 3 strengths and 3 weaknesses. For each one, quote 1 short phrase from the student’s text as evidence (max 12 words per quote). If evidence is unclear, say so.”
This habit discourages vague feedback like “good structure” with no explanation.
4. Separate “assessment notes” from “student-facing feedback”
GenAI can help produce two useful versions:
- internal notes (more detailed, includes uncertainties)
- student feedback (short, kind, actionable)
This is especially helpful when you want to be honest internally but keep feedback clean and motivating.
Example prompt:
“First write internal assessment notes (bullet points). Then write a student-facing feedback paragraph (80–120 words).”
5, Use GenAI to check completeness and alignment
Another safe use is “quality control”. GenAI can check whether feedback covers all rubric criteria, whether the tone matches your style, and whether the comment includes clear next steps.
Example prompt (quality check):
“Review the feedback text below. Does it (a) match the rubric, (b) include at least one concrete next step, (c) avoid vague statements? Suggest improvements.”
6. Support consistency across a teaching team
If multiple instructors grade the same assignment, GenAI can help keep feedback consistent – by using a shared rubric, shared wording style, and shared templates. This is one of the best uses of GenAI: reducing random variation.
Tip: create a “feedback template library” (common strengths, common issues, next-step suggestions) and let GenAI adapt it to each student.
Example prompt (team template style):
“Use the department feedback style: short sentences, neutral tone, focus on learning. Use this template set: [paste]. Adapt it to the student work summary.”
7. Common mistakes to avoid
- Asking GenAI to grade without a rubric (“Give a grade from 1–5”)
- Letting GenAI decide borderline cases without human judgement
- Copying GenAI feedback without checking accuracy and tone
- Giving feedback that is too generic because the prompt was vague
- Using GenAI in ways that violate privacy or institutional rules
8. Data protection and academic integrity
Before using GenAI with student work, consider privacy rules and tool policy. If the tool is not approved for sensitive data, avoid pasting full student submissions. Safer options include anonymising, summarising, or using only short excerpts.
Also consider transparency. Students should know whether GenAI was used for drafting feedback, and that the teacher is responsible for the final assessment. A simple statement can help: “GenAI may be used to draft feedback; final grading decisions are made by the instructor.”
9. A simple, safe “GenAI-assisted grading” routine
- Prepare a rubric (clear criteria)
- Skim the work and write quick teacher notes (your judgement first)
- Ask GenAI to draft feedback based on your notes and the rubric
- Ask GenAI to check for rubric coverage and actionable suggestions
- Edit, verify, and finalise (teacher remains accountable)
GenAI can make assessment faster and more consistent – but only when it is guided by clear criteria and careful review. Used this way, it supports better learning feedback, not just faster grading.