AI and Marking Disclaimer
Effective date: [to be set at launch] · Version: 0.5 (draft for legal review)
Short forms
On question papers:
Practice paper created with AI from your own study material. Not an official board paper.
On corrected papers and results, until a mentor approves them:
AI-suggested marks: awaiting mentor review. Marks, percentages, correct answers and explanations were produced by artificial intelligence and may contain mistakes. A mentor (teacher, parent or tutor) should review and approve them. Not an official result.
On corrected papers after approval:
Marks suggested by AI and reviewed and approved by [mentor name] on [date]. Not an official board result.
Full text
1. What the app does, and what it doesn't. Munarish SmartAssess uses artificial intelligence (AI) to read your study material, write practice questions, read handwritten answers, and suggest marks, percentages, correct answers and explanations. It is a learning and practice tool. It is not an examiner, it does not award official marks, and its results are not official results.
2. AI marks are suggestions until a mentor approves them. AI can be wrong. It can:
- misread handwriting (especially untidy writing, diagrams, symbols or crossed-out work);
- misunderstand an answer written in an unusual but correct way;
- give too many or too few marks, or a wrong percentage;
- write a "correct answer" or an explanation that is itself wrong or incomplete;
- point to the wrong part of your material.
That is why every result is marked "AI-suggested" until a mentor reviews and approves it. A mentor is a responsible adult who knows the subject: a teacher, tutor or parent (or, for adult learners studying on their own, the learner). In a school or academy, a teacher must approve results before students or parents see them. At home, the parent or the adult learner reviews and approves results in the app. Please don't rely on marks that no mentor has reviewed.
3. How to review. The app helps the mentor check quickly:
- it shows what it read from the answer sheet before marking, so reading mistakes can be fixed;
- each mark shows its reason, the percentage of the answer judged correct, the **correct answer and what was wrong or missing**;
- each judgement gives a reference to your own material: document, page and line numbers, with the quoted passage. Open the reference and check it. A reference the app couldn't confirm is shown as "reference not confirmed";
- marks the AI is less sure about are flagged "please check";
- the mentor can change any mark (with a reason), ask for a second look at a higher level, or approve the paper. Changes are shown as "adjusted by mentor".
4. What the percentage means. The percentage shown for each answer is the share of that question's marks the AI suggests the answer earned, based on the key points in the marking scheme you chose (for example, 3 of 5 key points = 60%). It is not a measure of how confident the AI is, and it is not a grade or rank. It may change when a mentor reviews it.
5. Only as good as the material and the rules. Questions, answers and references come from the material in your knowledge bank and the marking rules you set. If the material is incomplete, out of date or wrong, or the rules don't match your exam, the results will reflect that. If you turn on "Allow general knowledge", some questions may come from outside your material. They are labelled, and are more likely to differ from what was taught.
6. Practice, not official. Papers follow the general style of a board's or university's papers but are not official papers or predictions of exam questions. Munarish is not affiliated with or endorsed by CBSE, CISCE, Cambridge Assessment International Education, any State Board, university or examination body.
7. Not for high-stakes decisions. Results, percentages and progress reports are meant to guide practice and revision. They must not be used on their own for admission, promotion, grading, scholarships or any decision about a learner. Any such use needs a qualified teacher's own assessment.
8. Different levels, different results. Standard, Advanced and Expert may judge the same answer slightly differently. Every level has passed our quality tests for the tasks it is offered for, but none is perfect.
9. Report problems. If a question, mark, correct answer, explanation or reference is wrong or inappropriate, use Report in the app. We review reports and use them to improve the app. If a review shows the AI was wrong, any papers used for the re-mark are given back.
10. Your responsibility. By using the app you accept that AI suggestions may be wrong, and that you (or your mentor) are responsible for reviewing them before relying on them. Our terms of use (section 7 and 11) explain our responsibilities and their limits.