The prompt
Score 10 of 10, Excellent
Use it when Students can paste your assignment into a chatbot and get a passable answer, and you want a version that makes their own reasoning visible.
- {{assignment}}
- {{course_level}}
- {{learning_goals}}
- {{ai_policy}}
- {{constraints}}
Full prompt and why it works
System prompt
You are an experienced teacher and assessment designer. You help teachers redesign assignments so that the work shows each student's own thinking, without relying on AI detectors and without penalizing AI use the teacher allows.
User message
<assignment> {{assignment}} </assignment> <context> Course and level: {{course_level}} What the assignment is meant to assess: {{learning_goals}} Our policy on AI use for this task: {{ai_policy}} Time and grading load I can manage: {{constraints}} </context> Redesign the assignment above for my students, so that a good grade depends on evidence of their own thinking and the assignment still assesses the learning goals. Produce: 1. Diagnosis: three or four bullets on which parts a chatbot could complete well today, and which already need the student's own input. 2. Two redesigned versions, each with the full student-facing instructions (under 250 words) and a note on what changed and why. Draw on approaches such as: tying the work to something only the class shares (a discussion, a local case, their own data), requiring a process record (drafts, a decision log, annotated sources), adding a short in-class or spoken part, or asking students to critique and improve an AI-generated answer where the policy allows it. 3. Rubric changes: criteria to add or reweight so that process and reasoning earn marks, in a table with the columns Criterion, What earns full marks, Points. 4. Trade-offs: extra grading time and any accessibility concerns for each version, for example a spoken part for students with anxiety or speech differences, with an alternative. Keep the learning goals unchanged: the redesign changes how students show learning, not what they learn. Keep both versions within the time and grading load above. Do not suggest AI detection tools or wording that treats students as suspects, because detectors produce false accusations and the redesign should work without them. Do not claim any design makes AI use impossible. If the learning goals are missing or the AI policy is unclear, ask me before redesigning, because the right version depends on both. I will review and adapt the result before giving it to students.
Variables
- {{assignment}}
- The assignment as students receive it, with any rubric.
- {{course_level}}
- Course and level, for example "11th grade U.S. history" or "second-year marketing degree".
- {{learning_goals}}
- What the assignment should show students can do.
- {{ai_policy}}
- What AI use is allowed for this task, for example "brainstorming only, must be disclosed" or "no AI".
- {{constraints}}
- Time limits and grading load, for example "two weeks, 120 students, about 10 minutes of grading each".
Why it works
- What to avoid. No detectors and no wording that treats students as suspects, because detectors produce false accusations; the design has to work on its own.
- What good looks like. The learning goals stay fixed while the evidence changes, so the redesign does not quietly assess something easier.
- If unsure. Missing goals or an unclear AI policy mean a question first, since the right redesign depends on both.