Virginia Tech® home

Teaching

Using AI for Teaching

Guidance for instructors on four common places AI shows up in teaching: course policy, assignment design, feedback and grading, and tool adoption.

Getting Oriented

The instructor decides where AI fits in their course. That decision is shaped by several sources of information and guidance, including Virginia Tech’s Responsible and Ethical AI Principles, federal law, familiarity with AI tools, and AI literacy or training.

Responsible Use

In teaching, AI policies should be considered in the places where AI most commonly shows up: when setting course policy, designing assignments, giving feedback, and adopting new tools. Each recommendation below is followed by the related Principle for Responsible and Ethical AI in parentheses.

Setting your course AI policy

Adding a course AI policy to your course syllabus tells students what AI use is allowed, what is not, and what to disclose. Instructional support experts in Technology-enhanced Learning and Online Strategies (TLOS) and Center for Excellence in Teaching and Learning (CETL) recommend the following as a best practice for clarity.

  • The AI policy should align with the learning outcomes of your course. (Mission Alignment)
    • Consider what students must be able to do without AI to demonstrate their learning.
    • Write your policy to align with your discipline’s guidance on acceptable AI use.
  • Be explicit about what is allowed, what is not, and what students should disclose when they use AI. (Fairness & Transparency)
    • Use concrete statements, e.g., “AI may draft outlines; but you must write the prose," is clearer than “Use AI ethically.”
    • Consider asking for a one-line disclosure on any assignment where AI was used.
    • When AI expectations differ from one assignment to the next, state them in the assignment description so students see the rule at the moment they need it.
  • Make your AI use expectations clear. An ambiguous policy that students can’t confidently follow is hard to enforce later. (Responsible & Ethical Use)
    • Review the policy on the first day of class, with plenty of time scheduled to answer questions.
    • Include a list of approved tools that students can use in your course.
  • Preserve assignments where unassisted human work is the whole point. (Human Judgment)
    • Consider reserving at least one assessment for in-class work that is unassisted by AI.
    • Explain how the unassisted assignment helps students learn discipline-specific skills.

Designing assignments and activities

AI-aware assignment design asks for work AI cannot produce well, or work where a student’s own reasoning is obvious.

  • Assignments should align with learning outcomes, e.g., judgment, synthesis, domain expertise, and original voice. (Mission Alignment)
    • Ask students to compare, justify, or decide versus summarize.
    • Write assignments that ask students to respond to reference materials that aren’t likely to have trained AI tools, e.g., Virginia Tech data, a specific paper, or a current event.
  • Consider what assignments are better because students use AI or may help students learn skills they can apply in future employment. (Innovation for Good)
    • Include “critique-the-AI” assignments where students write prompts then analyze the output.
    • Require students to show their prompts, the output, and their edits.
  • Grade the student’s reasoning, not the polished output. (Human Judgment)
    • Grade the process, e.g., outlines, drafts, and revision notes as well as the final draft.
    • Assign short oral check-ins.
  • If you expect AI use, say so; if you forbid it, say that too. Ambiguous policies penalize students who try to follow them. (Fairness & Transparency)
    • Include your AI expectations in the assignment prompt as well as in the syllabus AI policy.
    • Explain how your AI expectations align with the course’s learning outcomes.

Giving feedback on student work

AI can speed up feedback by drafting language, flagging patterns, and suggesting edits, but care must be taken to protect privacy and fairness.

  • Feedback should help the student improve. (Human-Centered Benefit)
    • Review AI-generated feedback to ensure the comments are specific to the assignment and give actionable ways for the student to improve.
    • Remember that students are paying to receive the benefit of your expertise. Determine when your unassisted feedback provides your students the most benefit.
  • Student work is FERPA-protected. Confirm the tool you are using is approved by Virginia Tech for the appropriate data risk level before you upload. (Data Security & Privacy)
    • Sign in with your Virginia Tech credentials. Not a personal account.
    • Always anonymize or aggregate information, if possible.
  • Read the student’s work yourself before the AI analyzes it. (Human Judgment)
    • Form your own thoughts first to avoid allowing AI feedback to shape your opinion.
    • Avoid allowing AI to assign grades, even in an advisory capacity.
    • Confirm feedback is in your voice before it’s shared.
  • Disclose to students when you have used AI to assist you in providing feedback. (Fairness & Transparency)
    • Consider adding a line to your syllabus about how you plan to use AI in your grading workflow.
    • Draft and use an assignment-specific prompt when drafting feedback for all your students.

Adopting and evaluating new AI tools

New AI tools appear regularly. Evaluate them as you would a new textbook, with additional questions about where student data goes and how the tool fits the learning outcomes.

  • Know where student inputs go. Virginia Tech recommends against using tools that train on student data for required coursework. (Data Security & Privacy)
  • Adopt only tools that have the potential to meaningfully improve learning for your students. (Innovation for Good)
    • Pilot the tool before you require its use.
    • Consider how requiring the AI tool will help students learn skills they will use in future coursework and employment. 
  • Consider the tool’s cost. (Fairness & Transparency)
    • Can every student get access at no cost?
    • Has the tool been reviewed for accessibility? For example, can the tool be accessed by screen readers and keyboard controls?
    • For course assignments where AI use is not part of the learning outcome, be sure to plan for students who want to opt out of using AI.
  • You remain responsible for your course’s learning outcomes, regardless of which tool you adopt. (Human Judgment)
    • Test the tool with edge cases from your discipline. 
    • Teach students where the tool fails alongside the cases where it works well.

When to stop and get help

If you have questions about what responsible and ethical use looks like, consult with colleagues, academic leadership, or experts.

  • You are not sure how to write a prompt.
  • You are not sure where to start when adapting assessments to use AI.
  • A student challenges a grade that AI played a role in assigning. 
    • Consult your chair before you respond.
  • You’re considering using a tool you haven't confirmed is approved for use at Virginia Tech.
    • Maintaining data security is a shared responsibility, and it’s important for everyone to do their part.
  • AI feedback contradicts your assessment of a draft, and you can't explain why. 
    • Don’t share the feedback.
    • Consider if the AI tool is inserting bias into its feedback.
    • Rewrite the feedback to align with your professional expertise.
  • A student discloses AI use that is not clearly covered in your course’s AI policy. 
    • Pause the grade decision until you've added clarifying language for the next cohort.
  • Someone questions whether you are using AI fairly across all your students. 
    • Listen first to the complaint and consider if the AI tool is inserting bias into its feedback.
  • The tool you are using stops working, changes terms, or gets flagged for issues. 
    • Don't substitute something unverified mid-semester. 
    • Fall back to the unassisted feedback for the rest of the semester.
  • You have FERPA questions.
  • A student submits work that you suspect does not follow your course’s AI policy.

Virginia Tech Resources