NTLSN · Crash Course · Use with care

AI for marking & feedback, safely — a crash course

There's a safe, useful way to use AI around marking — and a lot of unsafe ones. Four short lessons on the line between them, then a self-check.

The one thing to remember: AI can help you prepare to mark — rubrics, exemplar feedback, consistency checks — but it does not grade, and identifiable student work is not yours to upload.
4 lessons~10 min read1 self-checkGrounded in assessment-integrity & data-privacy practice

The lessons

1
What AI can safely help withPreparation, not judgement
›

The safe uses are upstream of grading: building the tools and the consistency that make your marking better.

  • Draft and sharpen a rubric or marking guide.
  • Generate exemplar feedback phrases you then adapt.
  • Brainstorm common feedback for a whole-class debrief.
2
The hard linesWhere you must not go
›

Some uses are off-limits — for privacy, fairness and integrity. Hold these firmly.

  • Never upload identifiable student work to a public AI tool.
  • AI does not assign or decide grades — that's your academic judgement.
  • Follow your institution's assessment, privacy and AI policies, always.
3
Feedback drafting with AIYou review and own every word
›

AI can help you draft feedback faster — but only on de-identified text, and you remain responsible for accuracy, tone and fairness.

  • Work from de-identified excerpts, not named submissions.
  • Review and rewrite — every comment must be true and fair.
  • Keep feedback specific to the work, not generic AI filler.
4
Bias, fairness & verificationTrust nothing unchecked
›

AI can be biased, inconsistent and confidently wrong. Your fairness obligations don't transfer to a tool.

  • Check for bias and inconsistency; you are accountable, not the tool.
  • Verify any factual claim in feedback.
  • If in doubt, leave the tool out — fairness comes first.
◇ Bring it together — from the NTLSN commons
Marking & Moderation courseFeedback That Works courseAcademic Integrity courseFrameworks & policy

Before you use AI around marking — a quick self-check

✓
I use AI only to prepare — rubrics, exemplars, debriefs.
✓
I never upload identifiable student work to a public tool.
✓
AI does not assign grades — I do.
✓
I work from de-identified text and own every comment.
✓
I've checked for bias and verified factual claims.
✓
I follow my institution's assessment, privacy and AI policy.
Source & attribution. Curated from assessment-integrity and data-privacy good practice indexed by the NTLSN commons. This is practice guidance, not policy or legal advice. AI must not assign grades; never upload identifiable student work without explicit institutional clearance; your institution's assessment, privacy and AI policies govern.
A free NTLSN crash course · see them all at /crash-courses.html. Runs in your browser; nothing is stored or sent.  ·  ← Back to NTLSN