You stay in control by treating JeddAI as a drafting assistant, not a decision-maker: it proposes feedback and marking against your own rubric and success criteria, then you read, edit, override and approve every comment before a student sees it. The professional judgement remains yours.
What does 'human-in-the-loop' actually mean for AI marking?
Human-in-the-loop means a teacher reviews and approves every AI output before it counts, so the technology drafts and the professional decides. With JeddAI, the tool never marks a submission on its own or sends anything to a student without you. It reads student work and proposes feedback, and you remain the person who accepts, reshapes or discards each suggestion.
The distinction matters because feedback and grades carry real consequences for a learner. An AI model can misread context, miss a student’s history, or phrase something in a way that lands badly. Keeping a human in the loop means those judgements sit with someone who knows the class, the standard and the young person in front of them.
How does JeddAI keep you in the driver's seat by design?
JeddAI keeps you in control by design because it only works from the marking guides you give it. You connect or upload student work, and JeddAI drafts feedback and marking aligned to your own rubric, success criteria and comment banks. Nothing is generated from a generic template; the suggestions reflect the standards you have already defined.
From there, every draft is presented for your review. You edit comments in place, adjust a proposed grade, rewrite anything that does not fit, and only then approve. The workflow is deliberately a proposal-and-approval loop rather than an automated pipeline, which is what lets you save marking time without handing over the decision. Because approval is an explicit step, a draft only becomes real feedback once you have chosen to accept it, and until then it exists purely as a suggestion you can rework or set aside.
Why does teacher control matter for NCEA and moderation?
Teacher control matters for NCEA because internally assessed standards rely on your professional judgement, and those judgements are subject to NZQA moderation. If an AI tool set grades unchecked, you could not defend them as your own assessment decisions. Keeping yourself in the loop means the mark you record is one you have genuinely made against the achievement standard.
The same holds for feedback that feeds into learning. Whether a student is working towards an internal or an external assessment, the guidance they receive should reflect a teacher who understands where the standard sets Achieved, Merit and Excellence. Using AI to draft that feedback is fine; letting it replace your reading of the evidence is not. If you are ever asked to justify a decision, you want to be able to point to your own judgement, informed by the drafts rather than driven by them.
What should you check before approving AI-drafted feedback?
Before approving any AI-drafted feedback, check it against the actual submission and your own criteria, not just for whether it reads well. A confident, fluent comment can still misjudge the work, so a short and consistent review pass is what catches the few drafts that need reshaping and confirms the rest.
Where something is off, edit it in place rather than starting over, since usually only a sentence or two needs adjusting. Across a set of submissions this stays fast, because you are reviewing rather than writing from scratch, while still signing off on every comment personally.
- Accuracy: does the feedback match what the student actually wrote?
- Alignment: does it reference your rubric or success criteria correctly?
- Grade: is the proposed mark defensible against the standard?
- Tone: is it encouraging and appropriate for this student?
- Next steps: does it tell the learner what to do to improve?
How do you set JeddAI up so it reflects your standards?
You set JeddAI up to reflect your standards by giving it your marking guides before it drafts anything. Provide the rubric or success criteria you already use, and load the comment banks that carry your usual phrasing. The more clearly your expectations are captured up front, the closer the first draft lands to what you would have written.
This front-loading is where much of your control lives. A well-built rubric and a set of comment banks act as guardrails, so the AI is interpreting your framework rather than inventing its own. It also makes review faster, because drafts already speak in your language and against your standard. When your marking guides change, updating them once flows through to every future draft, so your control scales across a whole class or a whole term.
How do you build a review routine that saves time without losing control?
Build a review routine by working through drafts in a consistent order so nothing skips your judgement. Read the student’s work, then the drafted feedback, adjust tone and specifics, confirm the grade, and approve. Doing the same steps each time turns review into a quick, reliable habit rather than an open-ended task.
A steady routine is what makes the time saving real. Because JeddAI has handled the first draft, your effort goes into checking and refining, which is faster than composing every comment from a blank page. You keep the parts only a teacher can do, the professional call, and hand the repetitive drafting to the tool. If a draft and your own read of the work disagree, treat that as a prompt to look again rather than a verdict to accept, since the routine exists to surface those cases, not to wave them through.
Where does a tool like JeddAI fit in your marking workflow?
A tool like JeddAI fits as the drafting layer of your marking workflow, not the deciding one. It applies your rubric and comment banks consistently across a whole class, so similar work gets similar feedback, and it takes on the first pass so your time goes into reviewing and refining. The judgement, the grades and the final word stay with you.
If you want to save marking time while staying firmly in control, you can Get started with JeddAI and set it up against your own standards. Keep your review routine tight, edit freely, and let the tool handle the repetitive drafting so your professional judgement leads every piece of feedback a student receives.
| Stage | What JeddAI does | What the teacher decides |
|---|---|---|
| Setup | Reads the rubric, success criteria and comment banks you provide | Which standards and guides define good work |
| Drafting | Suggests feedback and a mark aligned to those guides | Whether to accept, reword or reject each suggestion |
| Review | Presents drafts for you to edit in place | Tone, accuracy and what each student most needs |
| Grades | Proposes a grade against your criteria | The final grade and any moderation notes |
| Sharing | Holds feedback until you approve it | When and what the student actually sees |
Frequently asked questions
Does JeddAI send feedback to students automatically?
No. Nothing reaches a student until you approve it, so you review and release feedback yourself.
Can JeddAI override my professional judgement on a grade?
No. It proposes a grade against your criteria, but the final grade is always yours to set.
Is using AI to draft feedback allowed under NCEA?
Teachers remain responsible for their assessment judgements. Check your school's assessment policy and current NZQA guidance on authenticity and the use of tools.
What if the AI's feedback doesn't sound like me?
Provide your own comment banks and edit drafts freely. Over time, the phrasing you keep shapes feedback that reads in your voice.
Do I still need to read every student's work?
Yes. Reviewing each draft against the actual submission is what keeps the loop trustworthy and your judgement sound.
Get started with Jeddle
Jeddle gives teachers and students instant, syllabus-aligned feedback powered by JeddAI.
Looking for study material? Browse Jeddle's Australian-English subject resources, or explore more articles on Using Jeddle.
