Instructors are experimenting with a variety of ways to integrate or mitigate use of GenAI tools in their courses, including in activities and assessments. This page offers several ideas around integrating and mitigating use. You can find additional ideas for assignments and activities, including a helpful decision tree for reviewing your course assignments, at the University of Michigan Course and Assignment(Re-)Design page.
Ideas for integrating genAI use
GenAI tools have a wide variety of use cases that may support a part of the learning process during activities and assessments, depending on course learning objectives. Some examples of uses instructors have experimented with:
- Assist research or brainstorming. Use CoPilot to generate a list of terms and concepts relevant to a particular text or topic. Students can use these terms as the first step in a multi-tiered research process. Which of the terms sparks your curiosity? Which of the terms seem essential to our understanding of this text, argument, or even their rhetorical and/or historical context? Which would you be interested in researching further and why? What types of sources do you think are best suited to defining or explaining a given term? In a supplemental step, students could also use CoPilot to generate definitions and/or explanations of these key terms. After doing so, students could conduct their own research to fact-check the AI chatbot answers and even compare/contrast findings.
- Engage analysis and critical thinking.
- Analyze how a generated text meets/does not meet specific criteria for success. Ask students to--individually or in groups--draft a thesis statement and submit it to CoPilot so that it uses the statement to generate a persuasive essay. Then, ask students to evaluate the essay according to the criteria for success in the rubric or transparent assignment by filling in the rubric or annotating the essay, referencing specific language from the rubric or assignment for support. Where does the essay meet or fall short of the criteria?
- Evaluate the AI chatbot's synthesis and analysis of existing texts they are familiar with. For example, the instructor can task CoPilot with generating a comparative essay that analyzes the gendered norms in “The Story of an Hour,” by Kate Chopin, and “The Yellow Wallpaper,” by Charlotte Perkins Gilman. Once the essay is generated, students can analyze its synthesis of the two stories and identify opportunities for improvement. Where might the essay’s argument be expanded or improved with more details from the stories or from researched support? In what ways is the essay’s argument or synthesis insightful and/or flawed?
- Contribute to revision. For example, students might submit a draft to the AI and ask it to generate an outline of the essay. With the outline in hand, students can evaluate the organization of the outline or their essay with the goal of identifying opportunities to clarify key points or expand particular paragraphs or lines of thought. Students might also use the outline as a discursive map that helps them plan where in their essay they could address potential counterarguments or include researched support.
Ideas for deterring genAI use
While GenAI can support student learning in specific contexts for specific goals, its use can also be counter to learning. Examples of ways instructors who don't want students to use GenAI for portions of or all of an assignment activity have tried limiting use include:
- Having students reference course-specific moments and materials such as lectures, discussions, labs, notes, handouts, or sources not otherwise available on the internet
- Having students respond to image-based or sound-based texts in your assignments (add alt text or captions for accessibility)
- Asking students to apply or connect personal experience and knowledge in relation to key concepts or topics (while genAI can generate personal-sounding content, inviting personal experience and knowledge can be an incentive for student engagement).
- Having students use social annotation tools such as Hypothes.is or Perusall to engage with texts.
- Chunking major assignments, such as essays, into multiple due dates for key steps, such as an outline, notes on sources, drafts, etc.