Has any student ever actually read what you wrote on a quiz?
Most teachers have thought it. Few say it out loud. According to Carl Hendrick, writing in 2025, roughly two-thirds of students either ignore written quiz feedback entirely or make revisions that produce no improvement. Jonathan Sim of Times Higher Education puts the review rate at around 30%.
It’s not a motivation problem. It’s structural. In Edcafe AI’s Dudley College case study, teachers using the quiz feature noticed the same pattern: students were more focused on their results than on the explanations beside them.
Once you understand why it happens, the question shifts. Not “why aren’t they reading it?” but what kind of quiz feedback actually gets used, and who should be writing it?
This post covers why quiz feedback gets skipped, what research found when AI feedback was tested against it, and a practical framework for splitting the work between AI and you.
Why students stop at the grade
The grade anchors everything
When a student gets a quiz back, the grade is the first thing they see. Once they have that number, their brain largely stops processing. The explanation of what went wrong arrives after the decision about how to feel has already been made.
A veteran English teacher, writing for English with Mrs. Lamp, puts it directly:
They look at the grade, feel some type of way about the grade, and then stuff the paper somewhere. 99% of them will never look at it again unless you make them.
That’s not a student failing to engage. That’s a student responding exactly as the system trained them to: the grade is the outcome, and everything else is noise.
Timing is the other problem
Feedback that arrives two or three days later lands in a different mental context. The lesson has moved on. The question that tripped a student up is no longer live in their mind. At that point, quiz feedback isn’t instruction. It’s a post-mortem on something they’ve already filed away.
Faculty Focus puts it clearly: feedback is most effective when delivered while the material is still active in a student’s mind. The window is short. Most written feedback misses it.
AI vs. Teacher Quiz Feedback: What Each Does Best
Neither type of quiz feedback is superior. They’re built for different moments in the learning cycle, and conflating them is where the frustration starts.
Research from EdTech Insiders draws a useful distinction between two types of feedback: Feed Forward (what to do next) and Feed Back (what specifically went wrong in a student’s reasoning). AI and teachers aren’t equally suited to both.
| AI Feedback | Teacher Feedback | |
|---|---|---|
| Type | Feed Forward | Feed Back |
| What it does | Tells students what to do next | Diagnoses what went wrong in their specific reasoning |
| Example | “Add a counterargument here” / “Review how photosynthesis differs from respiration” | “You’ve confused correlation with causation. Your example shows one event following another, not one causing the other.” |
| Why it works here | Consistent, scalable, arrives at the right moment | Requires knowing the student, the context, and the reasoning behind the error |
A 2025 study cited in EdTech Insiders found AI Feed Forward feedback statistically indistinguishable from teacher Feed Forward. But on Feed Back, teacher feedback scored significantly higher. AI is structurally weak at tracing a student’s specific reasoning error back to its source.
The takeaway isn’t that one is better. It’s that they do different jobs.
Here’s a deeper look at building quizzes that make both the Feed Forward and Feed Back layers work.
Read the guide → AI Quiz Making for Teachers: The Complete Guide
What two real classrooms found
Students engaged with the quizzes. They skipped the explanations.
In a six-month trial at Dudley College, tutors used Edcafe AI to run structured quizzes with GCSE resit students (students retaking a key UK qualification after not passing the first time, typically with a lot riding on the result). These were students coming in with a pass rate of around 22%.
They engaged with the quizzes. But the tutors noticed a consistent pattern: students were more focused on their results than on the explanations sitting right beside them.
Only one type of engagement actually predicted higher grades
A 2025 peer-reviewed study from King Fahd University compared students taught with AI tools, including Edcafe AI, against a group in traditional lectures. The AI-tools group showed stronger engagement across the board. But only one type of engagement predicted academic performance: behavioral. Not reading the feedback. Acting on it.
What AI feedback changes, and why students engage with it
It arrives before the grade does
The timing problem is structural, and AI feedback addresses it directly. Students get an explanation on the question they just answered, while the reasoning is still live in their mind. They haven’t moved on yet. They haven’t filed the quiz away. That’s the window quiz feedback actually needs to land in.
Most written teacher feedback arrives after that window has closed.
It doesn’t feel like a correction
A 2025 randomized study by Weidlich et al. found that students rated teacher feedback as less fair and harder to accept than AI feedback. They also showed less willingness to revise after receiving teacher feedback.
That’s not a criticism of teachers. It’s how students process correction from someone with authority over their grade. AI feedback reads as information. Teacher feedback, even when carefully worded, can read as judgment. The medium changes the emotional response.
At King’s School Chester, after 18 months of AI feedback, 91.2% of students agreed AI feedback was clear and understandable.
Try this before your next quiz
Before your next quiz, write a short explanation for each question while you’re building it. Two to three sentences on why the correct answer is right and where the wrong ones go astray.
In Edcafe AI, if you’re not sure what to write, the AI can draft it for you. The explanation is fully editable: you can tweak the wording, add subject-specific context, or rewrite it entirely, then move on.
Either way, when students go through the quiz, they see that explanation right after each answer, while the question is still live in their mind. Not two days later in a returned paper.
That’s the shift: quiz feedback that used to take an hour to write across 30 papers now gets done once, at the quiz-building stage, and reaches every student at the exact moment they need it.
In Edcafe, you can choose how students experience feedback, after each question, or after the full quiz. You can read more on how that works in Assign Quizzes in Three Modes with Edcafe AI
Google Forms works too, with per-answer feedback added under each question’s answer key. The tool matters less than the timing.
Tip: Keep quiz feedback short and action-oriented. "This answer confuses X with Y. The key difference is Z" works better than a paragraph. Students are more likely to read two sentences than six.
What to keep for yourself
AI handles the immediate layer well. What it can’t do is know why that error keeps showing up, or be the person the student needs to hear it from. No tool replaces that.
The feedback wasn’t wasted. The delivery system was the problem. That’s the trade worth making.