Faculty quick-start guides
Step-by-step guides to integrate AI into your teaching. Each takes under 30 minutes and requires no technical background.
← Back to Teaching with AIWrite your AI syllabus statement
~15 minutesA clear AI policy removes ambiguity for both you and your students. Research from Stanford's Center for Teaching and Learning shows that students respond better to AI policies when they understand the pedagogical reasoning, not just the rules.
Choose your integration level
Refer to the AI Integration Spectrum on the main page. Decide where each major assessment falls. Most faculty find a mixed approach works best: some assignments restrict AI, others allow guided or collaborative use.
Tip: You don't need a single blanket policy. Many universities now recommend assignment-level AI policies. A general syllabus statement sets the tone while specific instructions accompany each assignment.
Select and customize a template
Below are five adaptable templates drawn from policies at UT Austin, University of Michigan, University of Minnesota, Tufts, and Harvard. Choose the closest to your approach and customize the bracketed sections.
This course requires all submitted work to be your own, produced without the assistance of generative AI tools (including ChatGPT, Claude, Gemini, Copilot, and similar). This policy exists because [state your learning rationale, e.g., "developing your independent analytical voice is a core learning objective"]. If you are unsure whether a specific tool falls under this policy, please ask before using it.
Adapted from UT Austin CTL & University of Michigan GenAI guidelinesIn this course, generative AI tools may be used for [specify: brainstorming, outlining, grammar checking] but not for [specify: drafting, analysis, code generation]. Each assignment will specify whether and how AI may be used. When you use AI, you must include an acknowledgment statement describing the tool, how you used it, and what you changed.
Adapted from University of Minnesota GenAI syllabus statementsYou are encouraged to use generative AI tools as part of your learning process. When you use AI, you must: (1) disclose the tool and how you used it, (2) include the relevant prompts or conversation excerpts as an appendix, and (3) clearly indicate which portions of your work are AI-assisted vs. your own original contribution.
Adapted from Tufts University CELT & Harvard University guidelinesThis course takes a differentiated approach to AI. Some assignments will prohibit AI use to ensure you develop [specific skills], while others will require you to use AI and critically evaluate its outputs. Each assignment will clearly state its AI policy using the following labels: [No AI / AI Assisted / AI Collaborative]. When in doubt, ask before submitting.
Adapted from the AI Assessment Scale (Perkins et al., 2024)Generative AI tools are part of our professional landscape. You may use AI for any assignment unless explicitly restricted. All AI-assisted work must include a reflection paragraph explaining: what you prompted, what the AI produced, what you changed, and why. Your grade reflects your judgment and ability to improve upon AI output, not the AI's output itself.
Adapted from Michigan & the Sentient Syllabus Project (Boris Steipe)Communicate and discuss
Add your statement to your syllabus, then dedicate 10 minutes on the first day to explain your reasoning. Students are more likely to follow a policy when they understand the learning rationale behind it.
- Explain why you chose this policy for this course specifically
- Show an example of acceptable vs. unacceptable AI use
- Invite questions and make clear that asking is always safe
- Revisit the policy mid-semester as both you and students gain experience
Design an AI-resilient assignment
~20 minutesAn AI-resilient assignment isn't one that AI can't touch. It's one where the learning happens regardless of whether AI is involved. The goal is to design assessments where the process of thinking, not just the final product, is what you evaluate.
Pick one assignment to redesign
Choose an assignment you suspect is vulnerable to AI completion (typically a take-home essay, problem set, or report). Resist the urge to overhaul your whole course. Redesigning one assignment well teaches you more than revising five superficially.
Test it with AI
Copy your assignment prompt into ChatGPT or Claude. Read the output carefully. Would this pass? Would it earn a B? An A? Note what the AI does well and where it falls short. This exercise reveals what your assignment is actually measuring.
Tip: Try multiple AI tools and different prompting strategies. If the first attempt earns a C but a refined prompt earns a B+, that tells you something important about your assessment criteria.
Run the 5-point AI-resilience checklist
- Personal or local context: Does the assignment require students to draw on personal experience, class discussions, or local data that AI cannot access?
- Process visibility: Do students submit evidence of their process (drafts, outlines, AI logs, revision history)?
- Oral component: Is there an in-person element where students must explain or defend their work?
- Scaffolded stages: Is the assignment broken into checkpoints (proposal → outline → draft → final)?
- Higher-order tasks: Does the prompt require evaluation, synthesis, or original judgment rather than summary or recall?
Modify and update your rubric
Revise the assignment to address gaps from the checklist, then update your rubric. If you've added process documentation, allocate meaningful weight (15-25% is common).
Process Documentation (20%): Student submits at least two drafts showing meaningful revision, an AI-use log (if applicable), and a 200-word reflection on their decision-making process.
Re-test and iterate
Run your redesigned assignment through AI again. The goal isn't to make it AI-proof, but to ensure that even if a student uses AI, the assignment still requires genuine intellectual engagement.
Run an AI literacy class session
~30 min prep · 50 min sessionA single 50-minute session early in the semester can set expectations, reduce anxiety, and equip students with a shared vocabulary around AI. This lesson plan is discipline-agnostic and adaptable. It combines a brief framing segment, a hands-on activity, and a facilitated discussion.
Opening: frame the conversation (10 min)
Start by acknowledging reality: most students are already using AI tools. This session isn't about prohibition; it's about developing judgment. Share your course AI policy and the reasoning behind it. Key points to cover:
- AI tools are trained on patterns in text, not knowledge or understanding
- AI can generate plausible-sounding content that is factually wrong (hallucination)
- Different courses will have different AI policies, and that's intentional
- The goal is developing judgment about when AI helps and when it hinders learning
This TED talk from computer scientist Yejin Choi provides an accessible, engaging framing of what AI can and cannot do, ideal for sparking class discussion:
Hands-on: test AI in your discipline (20 min)
Give students a short, focused task from your discipline and have them complete it both with and without AI. Suggested formats:
- Summarize a short article and compare your summary with AI's
- Ask AI to solve a problem from the course, then verify the solution
- Have AI generate a paragraph on a topic, then fact-check every claim
- Ask AI for 5 sources on a topic, then check whether they actually exist
Students work individually or in pairs for 15 minutes, then spend 5 minutes noting where AI excelled and where it failed. Firsthand experience is more persuasive than any lecture about AI limitations.
Discussion: when does AI help vs. hinder? (15 min)
1. What did AI do well? What did it get wrong? Were the errors obvious or subtle?
2. If you had submitted the AI's output as your own, would you have learned anything?
3. In what situations would using AI genuinely help you learn more? When would it prevent learning?
4. How would you feel if your doctor, lawyer, or engineer relied on AI the way some students use it for assignments?
Close by reiterating your course policy and inviting students to come to you with questions rather than guessing what's allowed.
Follow-up: assign a reflection (5 min)
Ask students to write a brief (200-word) reflection: What did you learn about AI today that surprised you? How will it affect how you use AI in this course? This can be ungraded or count as participation.
Tip: The Harvard AI Pedagogy Project offers a free interactive tutorial that walks students through using LLMs responsibly. You can assign it as pre-work or use it during the session itself.
Create AI-enhanced feedback
~20 minutes to set upAI can generate a detailed first-pass of feedback on student work, which you then review, edit, and personalize. This does not replace your expertise; it gives you a starting draft to react to rather than a blank page. Faculty who adopt this workflow report spending less time on routine comments and more on substantive, individualized guidance.
Convert your rubric into a prompt template
Take your existing rubric criteria and rewrite them as instructions for an AI. Be specific about what to evaluate, what tone to use, and what level of detail you expect.
You are a university-level writing instructor. Evaluate the following student essay based on these criteria: 1. Thesis clarity: Is there a clear, arguable thesis? 2. Evidence quality: Specific, relevant evidence used? 3. Analysis depth: Does the student analyze, not just describe? 4. Organization: Logically structured with clear transitions? 5. Mechanics: Note recurring grammar or citation issues. For each criterion, provide 2-3 sentences of specific, constructive feedback. Use an encouraging but honest tone. Do not rewrite the student's work. End with one concrete suggestion for the most important area of improvement. [Paste student work here]
Generate, review, and personalize
Run the prompt with a student's submission. Read the AI's feedback critically: is it accurate? Does it match what you would say? Remove anything generic or wrong, add your own observations, and insert personal references to class discussions or the student's previous work.
Important: Always review AI feedback before sharing it with students. AI may miss nuance, misread student intent, or give advice that contradicts your pedagogy. Be mindful of student data privacy when using external AI tools; avoid uploading student names or identifiers.
Iterate your template
After using your prompt on 5-10 submissions, you'll notice patterns in what the AI gets right and where you consistently override it. Refine your prompt template accordingly. Most faculty find that after 2-3 rounds of refinement, the AI's first-pass feedback requires significantly less editing.
Build a prompt library for your course
~25 minutesA shared prompt library gives students structured starting points for using AI productively in your discipline. Rather than leaving students to figure out prompting on their own, you provide curated examples that model good practice and teach prompt engineering implicitly.
Identify 5-10 common course tasks
Think about recurring tasks in your course where a well-crafted AI prompt would genuinely help students learn. Common categories: understanding difficult concepts, brainstorming ideas, getting feedback on drafts, generating practice problems, and exploring alternative perspectives.
Craft and test each prompt
Write a prompt for each task, test it across 2-3 AI tools, and refine until the output is consistently useful. Good academic prompts specify the role, audience level, desired format, and constraints.
Explain [concept] as if you were a university professor talking to an undergraduate who has completed [prerequisite course]. Use one concrete analogy and one example from [discipline]. Keep it under 200 words. Then ask me one question to check whether I understood the core idea.
I'm a [year] student in [course name]. Review my draft paragraph below. Identify the strongest point in my argument and one specific weakness. Suggest how I could strengthen it without rewriting it for me. Be honest but constructive. [Paste draft paragraph]
Generate 5 practice problems on [topic] at the difficulty level of a university [course level] exam. Include a mix of conceptual and calculation-based questions. Provide answers separately so I can check my work after attempting each one.
Share as a course resource
Publish your prompt library on your LMS with clear usage guidelines. For each prompt, include a brief note explaining when to use it and what to watch out for (e.g., "AI may fabricate citations, always verify references"). Invite students to suggest improvements throughout the semester.
Tip: Consider making this a collaborative class resource. Students who discover effective prompts can submit them for your review. This builds the library over time and reinforces prompt engineering as a skill.
Audit your course for AI impact
~30 minutesBefore redesigning anything, understand how AI affects your course as it currently stands. This structured self-assessment walks you through each major assessment, evaluates its vulnerability, and produces a prioritized action plan.
List all major assessments
Create a simple table with every graded assessment, its weight, format, and your initial sense of how well AI could complete it.
Assessment · Weight · Format · AI Risk · Priority
Midterm Essay · 20% · Take-home · High · Redesign first
Lab Reports (×5) · 25% · In-lab + write-up · Medium · Add process docs
Group Presentation · 15% · In-class · Low · Minimal changes
Final Exam · 25% · In-class proctored · Very Low · No changes
Participation · 15% · In-class · Very Low · No changes
Test each assessment with AI
For each take-home or unproctored assessment, input the assignment prompt into an AI tool and evaluate the output. Rate each on a simple scale:
- High risk: AI produces work that would earn a B or above with minimal editing
- Medium risk: AI produces a passable draft but needs significant human work to reach a B
- Low risk: AI cannot meaningfully complete the task (in-person, hands-on, oral, etc.)
Create a priority action plan
Focus on high-risk, high-weight assessments first. For each one, choose an approach:
- Redesign: Modify using the AI-Resilience Checklist above
- Add safeguards: Keep the assignment but add process documentation, oral defense, or scaffolding
- Integrate AI: Build AI into the task so students work with it critically
- Accept the risk: For some low-weight assessments, the cost of redesign may not be worth it
Tip: You don't need to address everything in one semester. Prioritize 1-2 high-impact changes, implement them, gather data, and iterate. Incremental improvement over multiple semesters is more sustainable than a single overhaul.
Need personalized guidance?
Book a 30-minute consultation with the AI Hub team to discuss your specific courses, discipline, and goals.