Blog
End of term grading is the week when several classes hand in a longer paper at once. Finish a key and rubric a second reader can follow, collect and photograph the stack the day students sit, mark against that key, review every mark, then hand each exam back as a PDF or an email.
Grading end-of-term exams with AI follows that order. Graider is the AI red pen for paper grading, a product of DM Liquidity LLC. Teachers photograph the stack, grade on their own rubric or answer key, review and toggle feedback, keep the pen, and hand work back as a PDF or an email. Running that exam set lives on AI exam grading. This page is the plan for that week.
Your key. Their papers. You review every mark. You keep the pen.
Write the answer key and the rubric before the first student picks up a pen. A second reader, with the documents and a photograph, should be able to award the same points you would.
The answer key lists the responses you accept: the final value, an equivalent form, the short phrase, and the diagram label you will count as correct. When two answers are both right, both belong on the key, with the mark each one earns.
Mark allocation is the map of those points. If a question is worth 8 marks, say where they sit: two for a correct setup, four for the working, one for the final value, and one for the unit. The parts should add to the question total. Write that split on the key before the exam is copied.
Method marks are the points for a valid path, including a path that stops short of the final value. Name the method, the points, and one short example. If a slip in part (a) may carry into part (b), write the condition, the points, and whether the student must show the carried value. Treat a missing unit, a missing label, and a partial diagram the same way: name the feature, the points, and a one-line example.
Every scored row should read that way. For the longer walkthrough, see how to write a rubric AI can follow. Then read the finished key as if the script belonged to a class down the hall, and add a line anywhere you would have to guess.
Collect the stack in the order you will photograph it. Count the scripts against the roster before you leave the room, and note who was absent.
Photograph every page, including the back when a student continued there and any extra sheet stapled on. Use a flat pile, even light, and the full sheet in the frame. If a page is faint, or a hand covers a line, photograph it again before the student leaves.
Label the set by class. Two sections of the same exam stay as two stacks until the key is locked. Write any new decision onto the key, then mark the second class against that same document. Two printed versions need two keys.
Grading end-of-term exams with AI, in this plan, means the marks come from your documents. You upload the exam students sat, the answer key, and the rubric with its mark allocation and method marks. Graider marks the photographs against that key and rubric.
You review every mark. Open the script, compare the score to the row on your key, and override any mark your judgment disagrees with. You keep the pen on that override. Toggle feedback on when the comment should go home with the paper. Toggle feedback off when you will teach the item in class.
Work one class through, or one question across the set if that is how you already moderate. When a method mark needs a new line, write it on the key first, then apply it to the scripts you have already opened, so the rest of the pile uses the updated rule.
Hand each exam back as a PDF or an email, per student. The PDF is the marked paper you can print or file. The email is the same result sent to that student. Students see the exam after you approve the marks and choose the feedback. Release waits until you send it. Keep a PDF for the student who was absent. The comment they see is the one you toggled on, and a script you overrode shows the score you chose.
When two sections sit the same exam, they share one answer key, one mark allocation, and one set of method-mark rules. A colleague marking the other section should be able to award the ordinary cases from those documents.
After the first pass, sample the same question in both classes: a full-mark script, a method-mark script, and a script that missed the question. If the scores disagree, tighten a line on the key, then pass back through the scripts you already reviewed. You still review every mark in both classes.
Borderline totals, a method still missing from the key, and a page you cannot read stay with you. If the case will appear again, write the decision on the key. If it belongs to one script, leave it as an override and keep the pen on that mark.
A realistic exam week finishes the documents before the sitting and leaves room for review afterwards.
Finish the key, the mark allocation, and the method-mark examples the week the exam is printed.
On the sitting day, collect and photograph the stack, count it against the roster, and retake any faint page.
Review every marked script before release. When you are ready, hand back a PDF for each student, or an email per student.
The order stays the same: key first, photographs the day of the exam, review every mark, then handback. That loop is the one in grade against your own rubric. End of term grading is the longer version of that loop: more points per script, and a key you will defend in moderation.
Start while the exam is still a draft. Write the answer key, the mark allocation, and the method marks before the paper is copied. The sitting day is for collecting scripts and photographing the stack.
The responses you accept, the points for each part, method marks with a short example, and any follow-through you allow. A second reader should be able to award the same points from the key and the photograph.
You upload your own rubric and answer key. Graider marks the photographs against those documents. You review every mark, override any score you disagree with, toggle feedback on or off, and keep the pen.
Use one key and one set of method-mark rules for both classes. Sample the same question in each class. If a score disagrees, update the key and review those scripts again. You still review every mark.
As a PDF or an email per student, after you have reviewed the marks and chosen the feedback. Release waits until you send it. You keep the pen on any override.
Read the scores, toggle feedback on or off, and keep the pen on any override. Hand each exam back as a PDF or an email.