Summarize Assignments rubric feedback with Copilot

  • 6 minutes

After grading an assignment with a rubric, educators can use Copilot to transform their feedback into clear, actionable summaries for learners. This feature extends educator input rather than replacing it—Copilot always uses the educator's rubric selections and comments as the basis for suggested feedback.

How AI feedback suggestions work

In the Assignments Speedgrader, Copilot offers two ways to generate feedback summaries:

  • Summarize rubric as feedback: After selecting rubric values that reflect the learner's performance, Copilot can summarize those selections into readable feedback. If the educator also added rubric comments or wrote summarized feedback, Copilot incorporates that input as well.
  • Enhance feedback: Educators who write at least 50 characters of feedback can have Copilot expand and refine it. This feature is useful for turning quick notes into polished comments.

In both cases, the educator reviews, adjusts, and approves the generated content before learners receive it. Educators shouldn't assign grades based on AI-generated feedback.

Choose the language and tone for your feedback

Copilot can write feedback in three tones, which lets educators match the message to the learner's needs:

  • Basic: Clear and concise. Provides straightforward feedback without extra detail. This tone is ideal for quick corrections or clarifications.
  • Instructional: Actionable and detailed. Offers step-by-step guidance to help learners understand how to improve and build specific skills.
  • Coaching: Supportive and motivational. Encourages self-reflection, fosters growth, and inspires learners to take ownership of their learning journey.

Educators can select the tone after generating feedback and regenerate with a different tone if the first choice doesn't fit.

Use AI feedback in the grading workflow

  1. Open the assignment in Speedgrader.
  2. Review the learner's submission and select rubric values for each criterion. Optionally, add comments to individual rubric rows.
  3. Or write notes in the feedback field.
  4. Select the option to generate AI feedback (like Summarize rubric as feedback or Enhance feedback).
  5. Choose a language. Select ok.
  6. Review the suggested feedback. Edit as needed to ensure it accurately reflects the assessment, either manually or by using the AI-assisted free form field to describe any changes needed for the feedback.
  7. If needed, select a different tone (Basic, Instructional, or Coaching). The default tone is Basic.
  8. Confirm the feedback before returning the assignment to the learner.

Learners receive the finalized feedback when the educator returns the assignment. The AI-suggested text is never shared automatically; the educator must approve it.

Note

AI feedback suggestions are optimized for text-based assignments like essays and written projects. Results might be less effective for numeric or science-focused content.

Maintain educator judgment when using AI

Copilot generates suggestions based on the rubric, grade, language, and educator input. It doesn't evaluate learner work independently or assign grades. The educator is responsible for:

  • Assessing the quality of the submission
  • Selecting appropriate rubric levels
  • Reviewing and editing all AI-generated feedback
  • Deciding what feedback the learner receives

Copilot can save time on writing while preserving the educator's voice and professional judgment.

Tips for better results with AI-guided assignments

  • Use detailed rubrics: Clear criteria and descriptors give Copilot more context for generating relevant feedback.
  • Add comments while grading: Notes on specific strengths or areas for improvement help Copilot produce more personalized summaries.
  • Match tone to purpose:
    • Basic for simple corrections
    • Instructional for skill-building assignments
    • Coaching when learners need encouragement
  • Review before returning: Always read the AI-generated feedback to ensure accuracy and appropriate tone.

Tip

Use the thumbs up/thumbs down feedback in the grading interface to help Microsoft improve this feature.

Reflection: How might different feedback tones serve different learners or situations in your class? How does your knowledge of the context of your unit of study impact how you give feedback to learners?