Teaching Examples and AI Demos

These examples began as prompts for a Fall 2024 faculty session. They have been recast as reusable teaching patterns for Fall 2026: class preparation, hypotheticals, practice problems, course assistants, and critique of AI output.

Use the examples as starting points. Replace the course, readings, learning goals, and constraints with your own. For current tool access and approved platforms, use the AI Resources Portal. For syllabus language governing student use, use the AI Syllabus Guide.

Before uploading materials

Do not upload student work, grading materials, class recordings, exam questions, confidential facts, clinic/client material, or nonpublic course materials unless the tool is approved for that material and the use is authorized. When the question is tool access or data policy, check the AI Resources Portal rather than relying on a copied note in a prompt.

Level 1 — Common teaching tasks

Create images for slides

Use image generation only when the image advances the teaching goal. Give the model the doctrinal point, visual setting, and constraints.

These readings are for a class session in Intellectual Property at Penn Carey Law. Create a photo-realistic image that illustrates the factual setting of the Texaco fair-use dispute: a science laboratory with journal volumes on shelves and lab tables, and a researcher copying an article for colleagues. Do not include logos, real people, or legible copyrighted text.

Script or outline a class

This works best when you provide the assigned materials, course level, time block, and learning goals.

These readings are for an 80-minute class session in Intellectual Property. Create a class plan that covers the assigned cases in order, spends more time on Campbell and Warhol, includes questions for students, and leaves ten minutes for synthesis. Identify where students are likely to confuse the doctrine.

Create slides

Ask for slide structure and questions, not polished doctrine you have not checked.

Create a five-slide outline for teaching the Google fair-use case. Use questions rather than long bullet points. Each slide should have one teaching objective, one discussion question, and one caution about what students often misunderstand. Do not invent quotations or citations.

Create hypotheticals

Hypotheticals should test the uncertainty you want students to confront.

Give me three short hypotheticals for the end of class that test the theoretical tensions in fair-use doctrine after Warhol. Each should change one legally relevant fact and include a note explaining what issue the hypothetical is meant to surface.

Level 2 — Higher-value tasks

Create practice problems

Practice problems should include the answer conditions, not just the facts.

Create a practice problem based on this class session that students can answer in one to two pages. Make it hard enough that students must address ambiguity after Warhol. Then draft a model-answer outline that identifies the strongest arguments on each side and the facts that matter most.

Design a course assistant

A course assistant is a bounded AI helper grounded in your course materials and rules. The platform choice belongs in the AI Resources Portal; the teaching design questions are stable:

• What materials may the assistant use?
• What materials may students upload?
• What questions should the assistant refuse or redirect?
• Should the assistant give answers, hints, citations, or follow-up questions?
• How will students know the assistant does not replace the syllabus, assigned materials, or office hours?

Starter instruction:

You are a course assistant for Introduction to Intellectual Property. Answer student questions using the syllabus, assigned readings, and slides provided in this project. When the materials do not answer the question, say that clearly and suggest what the student should review. Do not provide legal advice. Do not write answers to graded assignments. Cite the relevant reading, slide, or class session when possible.

For more design guidance, see Building a Virtual TA for Your Course.

Critique AI output

AI-output critique is often better than AI-output production. Students learn more when they must find errors, omissions, and unsupported reasoning.

I am teaching an upper-level law class. Generate a flawed answer to this hypothetical that sounds polished but contains three substantive legal errors, one unsupported factual assumption, and one missing counterargument. Then create a teacher-facing key identifying each flaw. Do not include the key in the student version.

Status

Maintained for Penn Carey Law faculty. Examples may be revised as better teaching patterns emerge.