During Teaching Talks, faculty gather to discuss and debate ideas from a recent journal article or podcast episode (and make a dent in their TRL files and piles!). In Teaching Takeaway posts, we will share the key takeaways from the Teaching Talks discussion. These posts are a great way to get a pulse on ‘teaching thoughts’ at UMW.
Teaching Talk session: September 11, 2023
Article: “Using AI to Implement Effective Teaching Strategies in Classrooms” (Mollick & Mollick, 2023)
Resource Summary: Mollick and Mollick assert that AI can support implementation of evidence-based teaching practices requiring intensive faculty time commitments. The authors make cases for AI streamlining faculty workload related to five specific teaching practices:
- generating multiple examples and alternative explanations for concepts
- identifying and addressing student misconceptions
- using frequent low-stakes testing
- assessing student learning
- creating items for distributed practice exercises
The authors offer model AI prompts related to each teaching practice with relevant considerations for evaluating output. Can we anticipate a new era of evidence-based teaching using AI? Mollick and Mollick contend that we may be in the early stages of a renewed focus on using strategies that work in classrooms.
Key takeaways from the group discussion:
1. Faculty saw great potential for AI to streamline labor-intensive teaching practices like creating formative assessment items (e.g., for an in-class Jeopardy game or retrieval practice exercises later in the semester) or generating multiple explanations and examples for complex concepts. Assessment items or examples could be tailored to the specific student audience or unique classroom contexts in ways that test banks and online searches cannot replicate efficiently.
2. Prompt engineering (guiding AI output through careful layering and refinement of prompts) is a skill that both faculty and students need to learn to most effectively use AI. Before we can do it with students, we need to learn it ourselves. (Yes, opportunities are coming soon!)
3. Mollick & Mollick suggested using AI to evaluate themes in student understanding by asking AI to analyze student exit tickets or minute papers. Our faculty participants were much less comfortable with using AI in this context. Concerns cited included privacy of student work, ethical considerations about reviewing individual student submissions vs. a whole group analysis, and potential errors in analysis (would verifying themes just require you to read all the responses anyway?). In the end, the group understood the authors’ premise, but were less likely to use AI for this purpose.
4. The article contained detailed examples of how to build prompts supporting targeted teaching objectives (see takeaway #2). If you have only played informally with AI tools but want to learn how to use them in a focused way for class design, the article models offer a solid step-by-step process to begin learning about prompt engineering.
Reach out with any questions and we hope to see you at a future Teaching Talk!
