AI Homework Evaluation Center
Eighty-four submissions marked overnight. Nine need your judgement.
The engine read every submission for concept understanding, reasoning, originality and presentation, and wrote its evidence for each score. It has flagged the nine where its own confidence is too low to stand alone.
Marking runs against the rubric you published, never against a hidden scale. Every score carries the evidence that produced it.
Submission Review Queue
Every submission the engine marked, ordered by how much your judgement is needed rather than by score.
Liam Walker
Class IV-B · Mathematics
Algebra Set 14 · Linear Equations
Kabir Sethi
Class VIII-A · Physics
Motion & Force · Problem Sheet 7
George Kettut
Class II-A · English
Descriptive Writing · A Rainy Day
Sophia Reyes
Class VIII-A · Science
Photosynthesis · Lab Report
Aarav Menon
Class VI-B · Mathematics
Fractions · Word Problems
Maria Fernandes
Class IX-C · History
Industrial Revolution · Source Essay
Why the AI Marked It This Way
Six findings on Liam Walker’s Algebra Set 14, written as margin notes. Reasoning on the left, the evidence that produced it in the middle, what to do about it on the right.
Excellent concept understanding
Every question was set up correctly before any arithmetic began. Where the method was right and the answer wrong, the error was always in the final step, never in the approach. That separation is the clearest signal that the concept itself is secure.
Move to the extension problems for this unit; the standard set is no longer informative.
Learning impactReinforces the topic
Add to learning planIncomplete solution
Three of nine questions stop partway through. Each break occurs at the point where the working requires a second substitution, and each abandoned attempt was otherwise correct up to that step.
Set four questions that isolate the second substitution step and nothing else.
Learning impactBlocks the next unit
Add to learning planWeak logical explanation
Answers are correct but unexplained. Where the rubric asks the student to justify a step, the response gives the result again rather than the reason. The understanding is present; the articulation of it is not.
Ask for one sentence of reasoning per answer before the numerical result.
Learning impactAffects marks, not understanding
Add to learning planStrong creativity
Question 7 was solved by a method not taught in class. The approach is valid, shorter than the taught method, and the student showed why it works rather than simply asserting it.
Ask the student to present this method to the class.
Learning impactWorth recognising
Add to learning planGrammar improvements needed
Fourteen sentence-level errors across the written sections, concentrated in tense agreement. None of them obscure the mathematical meaning, so they carry no penalty under this rubric.
Attach the tense-agreement practice sheet as optional follow-up.
Learning impactPresentation only
Add to learning planPresentation can improve
Working is unlabelled and runs across the page rather than down it. Two answers were difficult to locate. The rubric awards four marks for presentation and this submission earned one.
Share the worked-example layout sheet and remark the presentation criterion only.
Learning impactCosts marks under the rubric
Add to learning planLearning Gap Center
Eight dimensions the rubric covers, each with what the engine found and what closing it would return.
Concept Understanding
LowSecure. Every question was set up correctly before calculation, across three consecutive sets.
Improvement suggestionMove to extension problems; the standard set no longer tells you anything new.
Expected outcomeSustains engagement and reveals the next real ceiling.
Problem Solving
CriticalThree of nine questions abandoned at the second substitution, each correct up to that point.
Improvement suggestionSet four questions that isolate the second substitution and nothing else.
Expected outcomeCompletion rises from 67% to an expected 95% within two sets.
Critical Thinking
LowSolved question 7 by an untaught method and justified why it works rather than asserting it.
Improvement suggestionAsk the student to present the alternative method to the class.
Expected outcomeConverts a private strength into a class resource.
Writing Quality
HighAnswers are correct but unexplained; justification is missing in six of nine responses.
Improvement suggestionRequire one sentence of reasoning before each numerical answer.
Expected outcomeRecovers the six reasoning marks lost under the current rubric.
Presentation
HighWorking runs across the page rather than down it and is unlabelled throughout.
Improvement suggestionShare the worked-example layout sheet and remark presentation only.
Expected outcomeRecovers three of four presentation marks with no new learning required.
Creativity
LowConsistently finds shorter routes when the taught method is long. This is a durable trait, not a one-off.
Improvement suggestionOffer the open-ended problem set as an optional extension.
Expected outcomeKeeps a capable student from coasting on routine work.
Time Management
HighSubmitted forty minutes late with the final three questions unfinished. The pattern repeats on longer sets.
Improvement suggestionBreak the next long set into two shorter submissions with separate deadlines.
Expected outcomeRemoves the late-submission penalty and the unfinished tail.
Research Skills
MediumSources are used but not cited. Two claims in the written section come from outside the taught material.
Improvement suggestionAttach the citation guide and ask for a source line under each external claim.
Expected outcomePrepares the student for the source-based work next term.
Smart Feedback Center
Eight actions available now, ranked by what each returns against the time it costs you.
Provide Additional Practice
Four questions isolating the second substitution step, where every abandoned attempt broke down.
Assign Revision
Re-attempt the three unfinished questions with the working laid out down the page.
Recommend Video Lesson
The eight-minute substitution walkthrough, which matches the format this student responds to.
Schedule Teacher Discussion
Ten minutes on why the reasoning sentence matters, not on the mathematics itself.
Peer Learning
Pair with Kabir Sethi, whose presentation earned full marks on the same set.
Extra Reading Material
The tense-agreement sheet, offered as optional follow-up rather than as correction.
Challenge Assignment
The open-ended problem set, given the untaught method used on question 7.
Parent Follow-up
A short note on the late submission pattern, which has now repeated on three long sets.
Homework Quality Studio
The engine’s reading of the work overall, in words rather than percentages. The letter is the rubric band, not a mark.
Assignment Quality
Work is set up correctly and finished carefully where it is finished at all. The losses come from abandoned questions, not from poor work.
Submission Consistency
Twelve of fourteen sets submitted. The two misses and the three late submissions all fall on sets longer than eight questions.
Learning Progress
Moved two rubric bands on concept understanding across the term while the class average moved half a band.
Knowledge Retention
Correctly used a method taught six weeks earlier without prompting. Retention is well ahead of the cohort.
Subject Understanding
Secure across linear equations and substitution. The one soft area is simultaneous equations, seen only twice so far.
Improvement Areas
Presentation and written justification. Neither reflects understanding; both cost marks under the published rubric.
Strength Indicators
Finds shorter routes than the taught method and can explain why they work. This has now appeared on four separate sets.
Study Habits
Work begins late and finishes in a rush. The last third of every long set is measurably weaker than the first two thirds.
Teacher Copilot
Seven things the copilot can draft from this evaluation. Everything arrives as a draft for you to edit, never sent directly.
Evaluation Timeline
Where this submission sits between the day it was set and the day it closes.
Homework Assigned
24 Jul · 09:00Algebra Set 14 issued to Class VIII-A with the published rubric attached.
Submission Received
28 Jul · 22:40Received forty minutes after the deadline with three questions unfinished.
AI Evaluation Started
29 Jul · 06:15Engine read the submission against all five rubric criteria and recorded its evidence.
Teacher Review
31 Jul · PendingFlagged for judgement: originality confidence sits below the 85% threshold for auto-release.
Feedback Generated
PendingStudent, teacher and parent versions will be drafted once the review is confirmed.
Student Revision
PendingThree unfinished questions to be re-attempted with the working laid out down the page.
Evaluation Completed
PendingFinal marks recorded and the learning plan updated with the outcome.
Learning Opportunity Center
What could follow this evaluation. Each tile drops straight into the learning plan.
AI Insight Center
What the engine noticed across all eighty-four submissions that no single one would show.
Common Mistakes
Thirty-one of eighty-four submissions break at the second substitution. This is not eighty-four individual difficulties; it is one teaching gap appearing eighty-four times.
Top Performing Students
Sophia Reyes, Kabir Sethi and Nathan Cole scored above 88 on every criterion. All three laid their working down the page rather than across it.
Frequently Missed Topics
Simultaneous equations were attempted by 62 students and completed correctly by 19. No other topic in this set falls below 70% completion.
Strong Learning Patterns
Students who wrote a reasoning sentence before each answer scored 11 marks higher on average, whether or not the answer itself was correct.
Weak Concepts
Substitution into a second expression is the single weakest concept in the year group. It underlies four of the six most-missed questions.
Recommended Teaching Strategy
Reteach the second substitution as a standalone step before the next set, using the shorter method George Kettut arrived at independently.
Emerging Learning Trends
Late submissions have risen from 8% to 19% this term and cluster entirely on sets longer than eight questions. Set length, not difficulty, predicts lateness.