The "Rubric" Prompt
Key Takeaways
A clear rubric starts with a clear assessment, not a clever prompt. AI can help you draft the structure, but the instructor remains responsible for the expectations and the final decisions.
Define the learning goals and task before asking AI to draft a rubric.
Request observable criteria and distinct descriptions for each performance level.
Give the prompt context, scoring rules, format, tone, and accessibility constraints.
Check the draft for alignment, fairness, clarity, and accurate scoring.
Share the rubric early, use it consistently, and revise it when experience calls for a change.
What an AI rubric generator can—and can’t—do
A well-framed request can turn assignment instructions into a useful first draft, giving an instructor somewhere concrete to begin. It can also surface questions that are easy to miss when planning an assessment under time pressure. Still, a draft is not a decision: only the instructor can judge whether the criteria reflect what students were actually asked to learn and do.
Turn assignment instructions into draft criteria and performance levels
An AI rubric generator can organize information you provide into criteria and descriptions of different performance levels. The more specific the assignment instructions and learning goals, the more relevant the draft is likely to be. If the input says only “write a strong essay,” the output may fall back on broad labels such as clarity or quality; an assignment brief that defines the argument, evidence, and audience gives it something more concrete to work with.
Save time without replacing an instructor’s judgment
A draft may help with the repetitive work of putting criteria into a consistent structure, but it cannot decide what matters most in your course. An instructor still needs to determine whether a criterion deserves more weight, whether a standard is appropriate for the class, and whether the language treats students fairly. Keep the tool in the role of a drafting partner, not the authority on learning.
Identify details the prompt may need to clarify
If a generated rubric seems generic, treat that as a useful signal: the request may have left out important context. For example, “evaluate a presentation” does not say whether the emphasis is on research, delivery, visual design, or discussion with an audience. A follow-up prompt can ask the AI to list assumptions and questions before it revises the rubric, rather than filling the gaps with guesses.
Keep expectations transparent for students
A rubric is most useful when students can understand what the criteria mean before they begin. Explain unfamiliar terms and describe what evidence of learning looks like in the particular assignment. When expectations are visible early, students have a better chance to plan their work—and to ask a sensible question if a requirement is unclear.
Define the assessment before writing your prompt
The quality of a rubric prompt depends on the assessment details behind it. Before opening a chat, take a moment to name the intended learning, the task students will complete, and the conditions under which their work will be assessed. This small pause can prevent a polished-looking draft from measuring the wrong things.
State the assignment’s purpose and learning objectives
Start with what students should know or be able to do by the end of the assignment. Then distinguish those goals from the topic or format: a report may be the vehicle for demonstrating analysis, but “write a report” is not itself a learning objective. If your goals are still broad, narrow them to actions you can observe in student work, such as comparing evidence or explaining a process.
Describe the student task, format, and audience
Tell the AI what students will produce and who will read, watch, or use it. Specify whether the work is an individual essay, a group presentation, a prototype, or another format, along with any important requirements such as a word range or source expectations. A useful prompt-building framework is to give the request a goal, context, format, and constraints, rather than relying on a short instruction with several unstated assumptions (a prompt-building framework).
Specify grade level, subject, and relevant standards
A criterion that suits an advanced seminar may be inappropriate for a younger class, even if the assignment topic is similar. Give the grade or learner level, subject area, and any standards you want the draft to reflect; include the actual standard wording when precision matters. These details guide the tone and level of complexity, but they do not verify that a generated rubric is aligned, so check the result against the source material yourself.
Set the number of criteria and performance levels
Decide how much detail students and graders can use without losing sight of the assignment. A compact rubric might have a few broad criteria, while a complex project may need more specific rows. This planning table can help you choose a workable starting point; it is a guide, not a fixed rule.
Assessment feature | What to decide | Example question |
|---|---|---|
Criteria | Which learning goals belong in the rubric? | Does each row assess something students were asked to do? |
Performance levels | How many distinctions will help students act? | Can a learner tell the difference between adjacent levels? |
Scoring | How will the results be combined? | Do points reflect the relative importance of each goal? |
Detail | How much explanation is useful? | Can students apply the descriptions without guessing? |
Once you have these choices, include them in the prompt. A rubric does not become clearer simply by adding more rows or levels; the useful amount is the amount that makes expectations easier to understand and apply.
Build a prompt that produces a usable rubric
A strong prompt gives the AI a specific job, enough information to do it, and boundaries for the answer. Think of it as a short assignment brief for the tool. The steps below help you ask for a draft that is easier to inspect and revise, rather than a tidy table full of generalities.
Ask for observable, assignment-specific criteria
Ask for criteria that describe evidence visible in the student’s work. For example, “uses relevant evidence to support the claim” gives a grader something to look for, while “shows excellent understanding” needs further explanation. You can also ask the AI to check that each criterion relates directly to a stated learning objective and avoid adding requirements not found in the assignment.
Describe how performance differs at each level
Performance descriptions should clarify what changes from one level to the next, not repeat the same praise or criticism with different adjectives. Tell the AI to describe observable differences—for instance, whether evidence is relevant, explained, or connected to the claim. Then review each row: if two neighboring levels could be swapped without changing their meaning, ask for a more precise distinction.
Request clear, student-friendly language
A rubric should make sense to the people using it, including students who are new to the subject. Ask for plain language, short descriptions, and a brief explanation of technical terms that cannot be avoided. If your goal is to check the request itself for vague instructions or missing context, a guide to avoiding common prompt problems offers another useful lens.
Set the format, such as a table or point scale
Specify how you want the draft organized, such as a table with criteria in rows and performance levels in columns. If points are involved, state the scale and whether each criterion has equal weight. You can also ask the AI to separate comments about the rubric from the rubric itself, so suggestions do not accidentally appear as student-facing criteria.
A short video can be useful for seeing how educators approach rubric design and assessment planning; use it as a prompt for reflection, not as a substitute for checking your own course goals.
After you receive a draft, read the criteria as if you were a student preparing the assignment. If you cannot tell what successful work would include, the descriptions need another pass before you share them.
Start with a reusable rubric prompt template
A reusable template saves time without making every assessment look the same. Keep the structure steady, then replace the assignment details, goals, scoring approach, and constraints for each new task. The template is a starting point; it should leave room for the particulars that make one assignment different from another.
Fill in the assignment context and learning goals
Include the subject, learner level, purpose, task, format, audience, and learning objectives in plain language. If an objective is especially important, say so rather than expecting the tool to infer its priority. USchool curates expert knowledge into simple, step-by-step frameworks; that broad learning focus is separate from rubric generation, but it reflects the value of making complex material easier to work through.
Add criteria, performance levels, and scoring requirements
Name how many criteria and performance levels you want, what labels to use, and whether the rubric should assign points or weights. Then ask for each criterion to connect to one or more learning goals, and for the scoring method to be shown clearly. For a first draft, the following compact sequence can help keep the prompt complete:
Describe the assignment, its audience, and its purpose.
List the learning objectives and any required standards.
Specify criteria, performance levels, and scoring rules.
Request a format and language level students can use.
After filling in the sequence, read it once as an instructor and once as a student. If a key expectation is understandable only to you, add a short explanation before asking for the draft.
Include constraints for tone, length, and accessibility
Say whether the wording should be concise, encouraging, and appropriate for the students’ reading level. Include any relevant accessibility needs, such as avoiding unexplained jargon or relying on a visual cue without a text description. Keep the learning goals intact while adapting how expectations are communicated; a format adjustment should not quietly change what the assessment measures.
Ask the AI to flag assumptions or missing information
End the prompt by asking the AI to identify unclear instructions, assumptions it made, or information it still needs. You can tell it to list those questions before drafting, or to mark uncertain parts for your review. This is a practical safeguard: an unanswered question is easier to correct than a confident but unsupported criterion embedded in a finished-looking rubric.
Adapt the rubric to different assignments and learners
The same prompt structure can support very different tasks, but the criteria must follow the work students are asked to produce. A rubric for an essay should not be a presentation rubric with the word “essay” substituted, and an exam may call for a different kind of scoring altogether. Match both the assessment method and the level of detail to the learning goals.
Adjust criteria for essays, presentations, projects, or exams
For essays, criteria might address the argument, use of evidence, organization, and clarity when those elements are part of the assignment. For a presentation, the criteria may focus on explanation, organization, or delivery if those are explicit goals. Projects and exams call for their own evidence of learning. Assignments can also draw on a wide range of subject matter: students might plan around Menehune Shores, consider polished concrete planning, or study the Evel Knievel Experience. Those topics do not dictate the rubric; the actual task and its learning objectives do.
Choose points, percentages, or proficiency labels
Points can make the scoring calculation direct, percentages can clarify relative weight, and proficiency labels can keep attention on descriptions of learning rather than arithmetic. No one approach is automatically best for every class. Whatever you choose, explain how the scores or labels will be interpreted and ensure that the method matches your institution’s grading expectations.
Account for accommodations without lowering learning goals
When accommodations apply, consider whether the format of an instruction or response needs to change while keeping the intended learning objective clear. Avoid assuming what a student needs; follow the relevant educational plan and consult the appropriate support staff when needed. The rubric should assess the stated learning goal, not penalize a learner for a method of participation that has been accommodated.
Match rubric detail to the assignment’s stakes and complexity
A low-stakes practice task may need only a few clear criteria and brief descriptions. A longer, more complex project may need enough detail to guide planning, feedback, and scoring across several stages. Examples from other fields can make the range of possible tasks concrete: a lesson might address engineered wood sanding or basketball court friction testing, while the rubric still needs to focus on the specific knowledge or process the assignment asks students to demonstrate.
Review and refine the AI-generated rubric
A generated rubric can look finished before it has earned students’ trust. Review it in context, compare it with the original assignment, and check each part for clarity and fairness. The goal is not to polish the language for its own sake; it is to make sure the rubric guides and assesses the learning you intended.
Check alignment with learning objectives and instructions
Compare every criterion with the assignment brief and the learning objectives. Ask whether the rubric rewards the intended skills, whether it introduces an unannounced requirement, and whether it leaves out a goal that matters. If a criterion cannot be tied to a stated objective or instruction, revise it, remove it, or explain why it belongs.
Look for vague, overlapping, or unmeasurable criteria
Words such as “excellent,” “appropriate,” and “strong” can be useful labels, but they need observable descriptions to support consistent judgment. Look for criteria that measure the same thing twice, combine unrelated skills, or rely on a personal impression. A small revision that makes the evidence visible is often more helpful than adding another row.
Verify that performance levels are distinct and fair
Read the descriptions across each row and check that the levels represent meaningful differences in work. A student should be able to understand what improvement looks like, not just which label is higher. Consider whether the descriptions leave room for different approaches that still meet the learning goal; a rubric can be specific without requiring every student to produce identical work.
Confirm scoring math and explain any necessary revisions
Recalculate point totals and weights rather than trusting a table to add itself correctly. Check that the maximum score matches the grading plan and that no criterion receives unintended emphasis. If you change the rubric after students have begun, explain what changed and how the revision will affect their work or scoring.
Use the rubric consistently in teaching and feedback
A rubric does more than organize grading: shared expectations can shape how students plan and revise their work. Introduce it while students still have time to use it, then apply it consistently when reviewing submissions. A thoughtful process also includes looking back at whether the rubric helped students understand the task.
Share expectations with students before they begin
Give students the rubric alongside the assignment instructions, not only after grading. Walk through a few criteria and invite questions about words or expectations that could be interpreted in more than one way. If students can identify what a criterion asks them to do, they can use it as a planning tool rather than discovering its meaning after the work is complete.
Calibrate scoring with sample work or a colleague
If possible, score a sample response against the rubric before evaluating a full set of submissions. Compare interpretations with a colleague when multiple people will grade the work, and discuss any criterion that produces noticeably different judgments. Calibration does not remove professional judgment; it helps make that judgment more consistent and reveals unclear language early.
Give feedback tied to criteria and next steps
Connect comments to the criterion they address, then offer a specific action the student can take. Instead of writing only “needs more detail,” point to where evidence needs explanation or where the reasoning could connect more clearly to the claim. This makes feedback easier to act on and shows students how the rubric relates to their next attempt.
Revise the rubric after reviewing student results
Once the work is graded, look for patterns. If many students misunderstood the same criterion, the assignment or rubric may need clearer wording; if a level was rarely or never used, check whether its description was realistic and distinct. The next version can preserve what worked while responding to what the class showed you.
Conclusion
A useful rubric begins with clear learning goals, describes evidence students can recognize, and stays open to careful instructor review. Use AI to help organize a first draft, then make the professional decisions that keep expectations aligned, fair, and understandable. To explore learning materials and online courses, visit USchool’s eLearning resources, and choose resources that support your goals.
Frequently Asked Questions
What should I include in an AI rubric prompt?
Include the assignment purpose, student task, learner level, learning objectives, criteria, performance levels, scoring method, and preferred format. Add any tone or accessibility constraints, and ask the AI to flag missing information or assumptions.
Can AI write a complete rubric without assignment details?
It can produce a draft, but with little context the criteria may be generic or based on assumptions. Providing the assignment instructions and learning goals gives you a better starting point to review.
How many criteria should a rubric have?
There is no single ideal number. Use enough criteria to cover the important learning goals, but not so many that students and graders lose sight of the central expectations.
Should a rubric use points or proficiency levels?
Either approach can work. Points make numerical scoring explicit, while proficiency labels can foreground descriptions of learning. Choose a method that fits the assignment and explain how it will be applied.
How can I make rubric language clearer for students?
Use direct wording, explain necessary subject-specific terms, and describe observable evidence rather than relying on general praise. Ask a student or colleague to read the rubric and identify anything that could be interpreted in multiple ways.
How do I check whether performance levels are fair?
Compare the descriptions across each criterion to ensure they show meaningful differences and connect to the assignment’s learning goals. Check that the language allows more than one valid approach when the task permits it and does not introduce unannounced expectations.
When should I share a rubric with students?
Share it before students begin the assignment, while they can still use it to plan their work. Invite questions and clarify any criterion that students find difficult to interpret.



Comments