5 AI Prompts That Cut Your Marking Time in Half
Marking is different from report writing — you're not summing up a term, you're responding to one piece of work, over and over, twenty or thirty times in a row. That repetition is exactly where AI earns its keep, if you set it up properly first.
The trap most teachers fall into is pasting in a student's work and typing "give feedback." You get back generic, soft, AI-sounding comments that could apply to almost anyone. The fix isn't a better AI tool — it's a better prompt. Here are five that actually work.
1. The rubric-anchored prompt
"Here is my marking rubric: [paste rubric]. Here is a student's response: [paste work]. Identify which criteria are met, which are partially met, and which are not met. For each one, quote the specific part of the student's work that shows this. Don't soften or pad the feedback."
This forces the AI to ground its judgement in evidence from the actual work, not vague encouragement. You're checking its reasoning, not blindly accepting its verdict.
2. The "one strength, one fix" prompt
"Read this student response: [paste work]. Give me exactly one specific strength and one specific, actionable next step. The next step must be something the student can do in their very next piece of work — not a vague goal like 'add more detail.'"
This is the prompt to use when you've got 28 books and twenty minutes. It keeps feedback tight and forces specificity, which is the thing that actually moves students forward.
3. The misconception-spotter
"This is a Year [X] [subject] response: [paste work]. Identify any misconceptions or errors in reasoning — not spelling or grammar — and explain in one sentence each what the student seems to be misunderstanding."
Use this for maths working, science explanations, or essay arguments. It's faster than you reading line by line trying to diagnose why an answer is wrong, and it gives you the language to address it directly with the student.
4. The voice-match prompt
"Rewrite this feedback so it sounds like a teacher talking directly and warmly to a 13-year-old, not a formal report comment: [paste draft feedback]. Keep it under 3 sentences."
Run your own rough notes through this rather than asking AI to generate feedback from scratch. You did the thinking — AI just tidies the delivery. This keeps your voice in it, which students notice more than you'd think.
5. The batch-pattern prompt
"I'm marking a set of [X] student responses to the same task. Here are 5 of them: [paste 5 responses]. What are the common strengths and common errors across this group? I want to address these with the whole class before I keep marking."
Use this halfway through a stack, not at the start. Spotting a class-wide misconception early means you can address it in two minutes the next lesson instead of writing the same correction in feedback thirty times.
The non-negotiable
Never let AI assign a grade or mark on its own. Use it to draft, diagnose, and speed up the parts that are repetitive — then you make the final call on every comment and every grade. Your judgement of the student is the part AI can't replicate, and it's the part that actually matters to them.
If you only try one of these, make it #2. It's the fastest to use and the one most likely to change what your feedback actually looks like.
TeachSmarterAI helps New Zealand teachers use AI tools effectively. New posts every week.