Best practices for responsible AI use
AI tools in Microsoft Teams Assignments save time and improve consistency, but they work best when educators apply them thoughtfully.
Equity and accessibility
AI-generated content should serve all learners, not just some. When you're using Copilot for rubrics and feedback:
- Review with fresh eyes: AI-generated content might unintentionally reflect patterns that don't serve all learners equally. A quick review helps ensure that rubric criteria and feedback resonate across the class community.
- Use multilingual options: Copilot generates rubrics and feedback in multiple languages, which helps educators connect with multilingual learners and families.
- Keep standards consistent: AI suggestions work best when they reflect the same expectations for all learners, with adjustments to tone and scaffolding based on individual needs.
- Design for flexibility: Consider whether rubric criteria allow learners to demonstrate understanding in different ways and support the full range of learners.
Privacy considerations
Educators and learners should be aware of what information AI guide uses when it processes assignment data to generate feedback.
Copilot uses the rubric, grade, assignment instructions, and educator-provided feedback to generate suggestions.
- AI-suggested feedback is only visible to the educator until the educator approves the feedback and returns the assignment to the learner.
- AI guide doesn't process student submissions, retain information from assignments, or use learner data to train AI models.
Educators should follow their school's data privacy policies and be transparent with learners and families about how they use AI tools in assessments.
Refine your AI outputs
AI-generated content is a starting point, not a finished product. To ensure that outputs match educator intent:
- Provide detailed input: The more context Copilot has through rubric descriptions, assignment instructions, and grading comments, the more relevant its suggestions are.
- Review every output: Read AI-generated rubrics and feedback carefully before using them. Check for accuracy, appropriate tone, and alignment with learning objectives.
- Edit freely: Modify AI suggestions to reflect specific observations about a learner's work or to match a specific communication style.
- Regenerate when needed: If the first result doesn't fit, adjust the input or try a different tone and regenerate.
- Learn from patterns: Over time, notice which types of input produce the best results and refine your approach accordingly.
Be transparent with your learners
Consider how to communicate with learners about your use of AI in the assessment process.
- Let learners know when AI tools assist with rubric creation or feedback generation.
- Emphasize that the educator reviews and approves all feedback before it reaches learners.
- Model responsible AI use by discussing how these tools support—rather than replace—educator judgment.
Avoid common AI pitfalls
- Over-reliance on AI: Use Copilot to save time on routine tasks, not to replace careful assessment of learner work.
- Skipping review: Never return AI-generated feedback without reading it first. Errors or inappropriate suggestions can undermine trust.
- Ignoring context: AI doesn't know individual learners the way educators do. Add personal observations and adjust tone based on what each learner needs.
- Inconsistent application: Apply AI tools consistently across learners to maintain fairness. Don't use detailed AI feedback for some learners and brief comments for other learners without pedagogical reason.
Reflection: What would transparency about AI-assisted feedback look like in your class? How might you explain the educator's role in the AI assessment process to learners and families?