Tomorrow Ready Resources → Evidence of Thinking → Using AI for Feedback Without Losing Teacher Judgement
[P+S] Primary & Secondary Evidence of Thinking

Using AI for Feedback Without Losing Teacher Judgement

The win is not "AI marks for me." The win is "AI helps me communicate feedback more clearly, while I stay responsible for judgement and learning." Keep the roles straight from the start.
A Practical Three-Step Workflow
  • Step 1: use AI for feedback stems only, options like "Here's what I notice..." or "Next, try..." or "One way to deepen this is..."
  • Step 2: add the one sentence that names the student's specific thinking move, this cannot be generated without teacher context
  • Step 3: keep evidence and data inside school systems, do not paste identifiable student work into third-party tools

The teacher-written sentence is the non-negotiable step. It is what makes the feedback genuinely useful and what keeps teacher judgement at the centre of the process.

What the Teacher Sentence Does

AI can generate plausible feedback stems. It cannot generate the one sentence that names what this student did in this moment with this piece of work. That specificity is what makes feedback land. It is also what confirms to the student that a human who knows them made a professional judgement about their thinking.

Examples of the teacher sentence in practice:

  • "Your strongest thinking move here is how you connected your observation to the class experiment."
  • "The decision log shows you revised based on new information, which is the core of good inquiry."
  • "You made a disciplinary choice when you used the specialist term correctly in context."
Primary

Generate three feedback stems using AI, then add one teacher-written sentence that names the student's specific thinking move. That sentence is yours alone and cannot be generated well without your knowledge of the student.

Secondary

Use AI stems for structure, then add the specific observation: "The decision log shows you revised based on new information, which is the core of good inquiry." Specific, non-generatable, and pedagogically powerful.

Decision Checkpoint

What safeguards will you use to ensure privacy and accuracy: anonymising before using any tool, maintaining human oversight of every piece of feedback, and keeping assessed evidence inside school systems?

Governance Reminder

Check privacy expectations and your school policy before using AI in your feedback workflow. Ensure feedback remains accurate, unbiased, and equitable for students who need more targeted human support rather than AI-generated stems.

Print or save this resource as a PDF using your browser.

More resources in this theme

Browse all Evidence of Thinking resources →
← Previous: Privacy-First Classroom Routines for GenAI ← Back to Evidence of Thinking