AI Feedback Students Can Act On: An Answer to "Why Did I Lose Marks?"

The bell goes, the room empties, and one student hangs back with their laptop open. "Why did I lose marks on this?" The project says 11 out of 16 and "Good effort, keep it up." You marked forty of them over two evenings. You remember roughly why this one landed where it did, but nothing on the page says so. The student doesn't need encouragement. They need to know what to do differently next time.

That question is the best test of feedback there is. If the student can answer it themselves from what they were given, the feedback worked. This post walks through where EduGears AI gives students AI feedback they can act on, and where you stay in charge of it.

"Why did I lose marks?"

Feedback goes vague when time runs out, not when teachers stop caring. Specific comments take minutes per student. A class takes hours, and the fortieth comment is shorter than the first. By the time the work comes back, the student has moved on, and "good effort" gives them nothing to move on with.

What makes AI feedback for students usable

Feedback a student can act on tends to have four things:

• It's tied to the criteria. Not "good", but which level they reached on which criterion.
• It names a next step. One thing to improve beats a list of everything that was wrong.
• It arrives while it still matters, before the next piece of work, not after.
• There's a chance to use it: a retake, a resubmission, or a hint on the next problem.

Each of the tools below covers one or more of those.

Projects: the level on every criterion

A project shows its rubric up front, under How your work will be graded, so students know the criteria before they start. When they submit their write-up and files, the AI grades the work against that rubric criterion by criterion.

When the grade comes back, the student sees the rubric again, this time with the level they reached and the points awarded on each criterion. Below it are an overall comment, Strengths and Improvements. So "why did I lose marks?" has an answer on the page: this criterion, this level, and this to improve.

A project uses AI drafts, I approve. The AI's grade waits as a draft, you can change any criterion score and write the overall feedback yourself, and nothing posts until you approve it. There is more on setting projects up in AI-graded project assignments.

Uploaded work: the reason behind the score

For work you capture or upload yourself (a scanned page, a photographed worksheet, a recorded presentation), the AI Grading tool gives you more to work with. For every rubric criterion it proposes a level, points and a short comment. It also drafts overall feedback, strengths, areas to improve, and the reason behind the score.

Those are your working notes. You read the reasoning, adjust anything you disagree with, and write the Overall feedback (sent to gradebook). That note travels with the score to the student's LMS gradebook when you choose Save & post to gradebook. The AI saves you the drafting. What the student reads is what you chose to send.

Short answers: marks you can explain

Short answers are where "why did I lose marks?" comes up most, because a paraphrase can be right or wrong depending on the course. In EduGears AI quizzes, every short answer records how it was marked: by an exact match, by the AI, or without AI because the AI was unavailable. Each course can be set to Lenient or Strict marking. An answer the AI couldn't check waits for you instead of being marked wrong.

After submitting, students see each question's correct answer and explanation on the results screen. If you change a mark or leave a comment, that question shows Mark updated by your name and a Teacher feedback note with what you wrote, and the new score goes to the gradebook. The whole story is in AI marking you can see, set and correct.

Retakes that point at what to fix

Feedback is most useful when there's a next attempt to use it on. Turn on Each attempt builds on the last, and a retake opens on the same questions with the correct answers already filled in and the wrong ones left blank. Each one is labelled Incorrect last time, and Jump to next to fix takes the student straight to it.

The student spends the retake on exactly what they got wrong, without being shown their old wrong answer again. It's feedback turned into a task. Our post on calmer quizzes and retakes covers it in more detail.

Reflections: a note back, not a grade

Not all feedback should be a judgement. Reflections are never graded and never sent to the gradebook. If you turn on AI acknowledgement, each student gets A note back to you after they submit: a short, supportive response to what they actually wrote, not an assessment of it. For a student who has just admitted they found the module hard, being heard is the feedback that counts.

The AI Tutor: hints, not answers

Sometimes the most useful feedback is a nudge in the middle of the work. By default, the AI Tutor answers from your course's own materials, and it is told never to complete an assignment, quiz or test for the student. It explains, asks a guiding question or gives a hint instead. On a practice question the student hasn't attempted yet, it gives hints rather than the answer.

Students can choose how it teaches (step-by-step, Socratic and more), unless you set a style for the whole course. Either way, the thinking stays with the student. More in an AI tutor that sticks to your course.

Where you stay in the loop

• Uploaded work and projects: the AI drafts and you approve. No score reaches the gradebook until you approve it.
• Quiz answers: you can override any mark and leave a note the student sees.
• Patterns across the class: quiz analytics shows which questions caught the class out, down to the most common wrong answer, so feedback can go to the whole room too.

The aim isn't to hand feedback over to the AI. It's to get specific, criterion-level feedback to every student while the work still matters to them, with your judgement on top.