There's a version of AI in education that can produce a polished essay in seconds, creating one of the most common problems teachers now face.
Finished paragraphs arrive before the student has engaged with the content or the ideas. The argument sounds fluent, but the thinking behind it never truly belongs to the student. The learning the essay was meant to produce never happens.
A student should start with a claim that feels true, then run into trouble. For example, the evidence doesn't quite fit, or a counterargument exposes the weak point. As a result, the student revises, cuts, connects, and tries again.
That friction isn't a flaw in the process.
It is the process.
The wrong shortcut
Most debate about classroom AI gets trapped between two bad options:
- let students use anything; or
- ban every tool with AI in it.
But "AI" is too broad a category to make either rule useful.
A tool that writes an assessed response for a student is fundamentally different from one that reads the student's response and asks them to strengthen it. The former replaces student thought, creating clearly undesirable outcomes. The latter creates another opportunity for student practice.
How can these be put in the same category?
New Zealand's current education guidance draws the same line. Assessment evidence must be the student's own work. For external assessment, that's a hard rule: generative AI isn't permitted, and anything it produces can't be submitted as the candidate's work. Internally, the call sits with schools, who must set their own policy under that same principle.
This gives schools a simple test for any classroom tool: Who is still doing the thinking?
If the answer is not the student, the technology has solved the wrong problem.
Feedback without answers
Students do need help.
They need to know when an argument becomes descriptive, when evidence has been used but not analysed, and when an idea that feels "obvious" has not actually been explained.
What they do not need is an escape from putting in the work.
Good feedback points to the gap without quietly filling it in. It might say:
- Your evidence supports the theme, but how does the writer's choice create that meaning?
- This paragraph makes two claims. Which one is your argument actually built around?
- You have identified a technique. Now explain why its effect matters in relation to the question.
None of those prompts writes the next sentence. But all of them make the next sentence better.
This is the principle behind WriteWise. It never generates essays or rewrites student work. It analyses what the student has already written, aligns its feedback with the relevant NCEA standard, and identifies the decisions the student still has to make.
Keep the struggle. Remove the wait.
As alluded to in the introduction, the productive struggle of writing matters. However, the unproductive delay in receiving feedback does not matter.
When feedback arrives one or two weeks after an essay was written, the student has already mentally moved on. The reasoning behind their arguments is no longer fresh. Great feedback can easily become useless.
On the other hand, immediate feedback changes the rhythm:
write → receive feedback → revise → write again
That loop is where improvement lives.
WriteWise makes the loop faster without making the intellectual work disappear. Students receive specific annotations while they can still remember what they were trying to say. Teachers can then discuss feedback with a student, without the grunt work of reading hundreds of pages. Ultimately, the technology handles the first pass, while the teacher remains the professional in the room.
On the WriteWise platform, the typical student using feedback has improved by one full NCEA grade level. However, the most important part is how that improvement happened: it wasn't through a stronger essay being supplied — rather, it was a development in the students' own work.
What should schools do?
The question can no longer be:
"Is AI allowed?"
Rather, they should be asking:
- Does this tool preserve student authorship?
- Does it make students think more carefully, or let them avoid thinking?
- Is its feedback tied to the standard actually being assessed?
- Can teachers review, adjust and control how it is used?
- Does the student leave with a better piece of writing — and a better understanding of how they created it?
Those questions separate vital educational technology from answer generation.
The goal of teaching writing is clearly not to help someone produce a document using the fastest available method. It is to develop someone who can form an idea, test it, argue it, and communicate it with control.
AI can remove that experience. However, it can also hand it back faster and sharper, giving the student the opportunity for true development.
We built WriteWise for exactly this.