Elective · AI for Curriculum Development
Elective · AI for Curriculum Development
Design courses, build teaching materials, and create student resources with AI
1 hour on Zoom | Have your browser open next to Zoom, and a course or session you teach | Offered once a year — dates announced by school email
This is an elective, offered once a year by demand, and aimed at faculty and instructors who design courses and teaching materials. Dates are announced by school email and posted on the home page. It builds on the prompting framework (PTCF) from Core 1 and the custom GPTs and shared Projects from Core 3.
What You’ll Leave With
By the end of this hour, you will be able to:
- Use ChatGPT Edu to draft and refine course materials—learning objectives, session outlines, active-learning activities, and assessments
- Use Copilot Chat or ChatGPT Edu to turn an outline into a syllabus section, slide text, and a rubric you paste into Word, PowerPoint, and Excel
- Optional, if you already use them: use Claude Code, OpenAI Codex, or the Claude desktop app to build course pages as HTML you paste into Canvas
- Use Gemini Notebook (formerly NotebookLM — renamed July 2026; old links redirect) to turn your own readings into student study guides, quizzes, and audio overviews—grounded in your sources
- Apply backward design: start from learning objectives and let AI help build aligned activities and assessments
- Apply the teaching-specific ethics: academic integrity and disclosure, accuracy, accessibility, and student-data privacy (FERPA)
Run of Show
| Time | Segment |
|---|---|
| 0:00–0:08 | Overview: where AI fits in course design; which tools to use for what |
| 0:08–0:23 | Ethics & Governance: integrity, accuracy, accessibility, and FERPA |
| 0:23–0:42 | Activity 1: From objectives to an aligned session (ChatGPT Edu, Copilot Chat, or an agentic tool) |
| 0:42–0:57 | Activity 2: Turn your readings into student materials (Gemini Notebook) |
| 0:57–1:00 | Wrap-up and what to build next |
Overview (8 min)
Start from the objective, not the slide. Good course design works backward: decide what students should be able to do, then design the assessment that measures it, then build the activities and materials that get them there. AI is fast at drafting at every step—but you bring the disciplinary judgment about what is correct, current, and worth teaching.
The tools, and what each is best at.
- ChatGPT Edu (Drexel-approved) — open-ended drafting and refining: learning objectives, session outlines, case studies, discussion prompts, and rubrics. Build a custom GPT (from Core 3)—a “Lesson Planner” or “Rubric Builder”—once, and reuse and share it with colleagues. Use the current default model in the Drexel workspace.
- Copilot Chat (Drexel-approved) — the same kind of drafting, with output shaped for your Office files: a syllabus section to paste into Word, slide-by-slide text for PowerPoint, a rubric table for Excel. Drexel does not have the paid Microsoft 365 Copilot license, so there is no Copilot button inside Word or PowerPoint; you paste or upload, then copy the result back.
- Gemini Notebook (Google) — source-grounded: upload your own readings, lecture notes, or a chapter and generate study guides, quizzes, briefing docs, and Audio Overviews, all with inline citations to your sources. Excellent for student-facing materials—but a consumer/Google tool, so treat it as Low Risk Data only unless Drexel has approved it, and use public or non-sensitive sources. Confirm on Drexel’s AI Tools page.
- Agentic tools (optional) — Claude Code, OpenAI Codex, and the Claude desktop app can work with files in a folder: build a set of Canvas pages as HTML, check every link, and fix problems across many files at once. They are not Drexel-supported. Course materials with no student data (objectives, outlines, your own slides, public readings) are low-risk, so these tools are fine for curriculum work. Student data never goes in.
Objectives → assessment → activities → materials. Ask AI to help at each step, in that order. If you start by generating slides, you get content with no target. If you start from the objective, everything you build has something to align to—and something to check it against.
Ethics & Governance: Integrity, Accuracy, Accessibility, and Student Data (15 min)
The ethics thread runs through the core sessions—how AI fails and who’s accountable (Core 1), verifying and disclosing in daily work (Core 2), data and consent in shared use (Core 3)—and comes together in Core 4: Verify, Disclose, Decide. This elective applies that thread to teaching, which adds its own duties: you are handing materials to students who will trust them, and you are handling student data.
Model the integrity you ask of students. Set a clear AI policy for your course, and be transparent with students about where and how you used AI to build it. The standard you hold students to is more credible when your own use is open and intentional.
Verify everything you hand to students. AI invents facts, citations, statistics, and can misstate disciplinary content with confidence. Every figure, citation, and claim in a teaching material needs to be checked against the source—students will not catch errors you miss. The rule from Core 2 applies: if you can’t open it, you can’t cite it.
Design for accessibility and equity. AI can help here—alt text for images, plain-language versions, adjusting reading level, drafting captions. It can also bake in bias or produce materials that don’t meet accessibility standards. Check that examples and images are inclusive and that what you publish is accessible to every student.
Protect student data (FERPA). Protected student records—grades, rosters, identifiable student work—must not go into a tool that isn’t approved for that data. Grading or giving feedback on identifiable student work requires an approved tool and an approved use case; otherwise de-identify first. For Gemini Notebook and other consumer tools, use public or synthetic content only.
Four checks before it reaches students: Accurate? (every fact and citation verified against the source). Accessible? (readable, captioned, alt-text where needed). Inclusive? (examples and images represent your students). Compliant? (no protected student data went into an unapproved tool). If any answer is no, fix it before you hand it out.
Activity 1: From Objectives to an Aligned Session
Activity 1: From Objectives to an Aligned Session
Goal: Go from a learning objective to an aligned session plan, an activity, and a draft assessment—backward design, with AI doing the drafting.
Track A — ChatGPT Edu (drafting and refining)
Write one or two learning objectives for a session you teach (non-sensitive).
Use a PTCF prompt (from Core 1):
Persona: You are an instructional designer for graduate public health courses. Task: Draft a 50-minute session plan aligned to these objectives, with one active-learning activity and a short formative assessment (a check for understanding). Context: [your objectives; the audience; any constraints]. Format: Objectives, timed outline, activity instructions, and the assessment, in plain language.
Iterate: ask for a rubric for the assessment, a short case study, or three discussion questions. Push it: “Make this more active,” “Does the assessment actually measure objective 2?”
Track B — Copilot Chat (into your Office files)
- Open Copilot Chat, signed in with your Drexel account, and paste or upload your outline. Ask: “Format this as a syllabus section with weekly topics, readings, and due dates. Use headings I can paste into Word.”
- Ask for slides: “Write slide-by-slide text for a 10-slide deck from this outline: a title and 3 bullets per slide, plus speaker notes.” Paste it into PowerPoint, then use Design → Designer for layouts.
- Ask for a rubric: “Build a grading rubric for the assessment as a table: criteria in rows, four performance levels in columns.” Paste the table into Excel or Word.
Track C — Agentic tools (optional: Claude Code, Codex, or the Claude desktop app)
For people who already use one of these. Work in a folder that holds only course materials: no rosters, grades, or student work.
Build Canvas pages as HTML. Put your session plan in the folder and ask:
Turn session-plan.docx into three Canvas pages (overview, activity instructions, assessment) as separate HTML files. Use headings in order, accessible tables, alt text for any images, and no outside styles. Then check that every link works.
Paste them into Canvas yourself. In Canvas, edit the page, switch to the HTML editor, and paste. You stay in control of what goes live, and no access token is involved.
Advanced: the Canvas API, in a sandbox only. Letting the agent update a course through the Canvas API is fine only in a sandbox or practice course with no students and no student data. Keep your token in an environment variable on your own computer, never in the chat or a file the agent writes, and run a dry run first. Setting Up Your Course with AI shows how to create a token and write a safe script.
Check your work: open each page in Canvas’s student view. Do the headings, tables, and links still work after pasting?
Reflection
- Where did the AI’s draft need your disciplinary judgment to be correct or current?
- Did the assessment actually measure the objective—or just stay on the topic?
Activity 2: Turn Your Readings into Student Materials
Activity 2: Turn Your Readings into Student Materials
Goal: Turn your own sources into student-facing study materials that are grounded in—and cite back to—those sources.
Gemini Notebook is a Google tool. Treat it as Low Risk Data only unless Drexel has approved it for higher classifications. For this activity, use public or fully non-sensitive sources—an open-access article, your own published slides, a public report. No student data, no unpublished or sensitive material. Confirm the current status on Drexel’s AI Tools page.
Instructions
- In Gemini Notebook, create a notebook and upload one to three public sources (an open-access paper, your own slides, a public dataset summary).
- Generate a study guide and a set of quiz questions from the sources. Ask for a glossary of key terms.
- Generate an Audio Overview—a short, conversational summary students could listen to before class.
- Check every output against the sources. Gemini Notebook cites where each point came from; you confirm the accuracy, emphasis, and tone before anything reaches a student.
Discussion
- Which student material saved you the most time to produce?
- Where did grounding in your own sources beat a general chatbot’s answer?
Wrap-Up (3 min)
- Start from objectives; let AI help build aligned activities, assessments, and materials—you keep the disciplinary judgment.
- ChatGPT Edu for open drafting and reusable custom GPTs; Copilot Chat for content shaped for your Office files; Gemini Notebook for source-grounded student materials (public data only); agentic tools (optional) for building Canvas-ready HTML, never with student data.
- Teaching ethics: integrity and disclosure, verified accuracy, accessibility, and FERPA.
Try this before the term: pick one session you teach, write its objectives, and use ChatGPT Edu to draft an aligned activity and a quick formative assessment. Verify it, then teach it.
Confirm the approved data level for each tool on Drexel’s AI Tools page before putting real material into it. Never enter protected student records (grades, rosters, identifiable student work) into a tool that isn’t approved for that data. Gemini Notebook and other consumer tools: public or synthetic content only. Agentic tools (Claude Code, Codex, desktop apps): course materials only, and the Canvas API only in a sandbox course with no student data.
Give Feedback on This Workshop
Your feedback directly shapes how these workshops are revised and what gets scheduled next.
In the first question, select the session you attended.