This spring Extra Points will feature a series of conversations focused on how faculty and staff around the University of Minnesota are using generative AI to do University work.
Tell us about your roles in Chemistry and the Center for Learning Innovation at Rochester, and how those are informing your work with generative AI.
Tim Doherty: I'm a chemistry instructor at the University of Minnesota, Rochester. In our lab courses, we teach students how to write lab reports. With the advent of large language models, it became clear that this is something we can’t ignore. We wanted to explore how AI could help students write lab reports. Since it’s such an emerging area, there wasn’t exactly a blueprint for this—it’s something we had to figure out. That’s where the faculty fellowship came in; it gave me an opportunity to formally develop this approach.
Interviewers: Were students interested in using generative AI? How did it come into play?
TD: It started from my own interest in how AI would transform education. The timing worked out well. As I began integrating it into the curriculum, students were becoming more aware of and interested in using AI tools. If we had waited even one more semester, we might’ve been behind the curve. So it wasn’t driven by students initially but rather by our curiosity and belief that students need help using this technology.
Set the context for us: Tell us about your FFP GenAI project.
TD: The project aimed to help students learn how to write chemistry lab reports using AI. First, we had them write all the sections of lab reports without AI. Then, they rewrote them using generative AI to explore its usefulness, its limitations, and how to effectively prompt it. We focused on helping students critically evaluate the AI’s outputs rather than just accepting them at face value. For example, they learned how to refine prompts to improve results and where the AI struggled, such as in crafting solid discussion or conclusion sections.
Interviewers: As someone who’s not familiar with chemistry lab reports, how does generative AI come into play? What kind of input do students provide?
TD: That was one of the main challenges. For some sections, like the introduction, a simple prompt—“Write an introduction to an acid-base titration lab with a pH meter” will get students started, but they can easily find mistakes or omissions. This is a good example of how we approached learning to prompt better - they needed to be more specific with the prompts to get quality output. For procedures, we asked students to provide detailed bullet points from their actual lab work, which the AI could then convert into a properly formatted procedure. For discussions and conclusions, though, the AI’s usefulness was limited unless students were advanced users. They often found it easier to write those sections themselves.
Interviewers: It sounds like you scaffolded their use of AI. Did you eventually let them decide how to use it?
TD: Exactly. We started with structured practice: writing each section manually first, then using AI. For the final project, students were given two weeks to write a full lab report in any way they preferred. They could use AI as much or as little as they wanted and then reflect on its role in their process.
Interviewers: Do you have a sense of how many students opted to use AI for the final project?
TD: It varied by section, some pieces of the lab reports are more easily outsourced to AI currently. As part of the assignments we asked students to reflect on how much they used AI, and based on those reflections most students heavily relied on AI for introductions and procedures, but use dropped significantly for discussions and conclusions. For discussions and conclusions, about half of the students used AI, and they estimated that it contributed maybe 10% of their work.
Interviewers: Do you plan to use this approach again?
TD: Absolutely. AI is evolving rapidly, so we’ll need to refresh our materials, but I think it’s essential. People are using AI to write academic papers and more, so students need to understand how to use it effectively and ethically. Ignoring it isn’t an option.
Were there any specific ah-ha moments during this project?
TD: One key takeaway was how students self-regulate when they understand the learning objectives and see value in them. If we make the objectives clear and connect them to activities, students tend to use AI responsibly. It’s also important to help them distinguish between helpful and harmful uses of AI.
What were the students’ reactions to the project?
TD: Surprisingly positive. I expected some pushback about not using AI for the first four reports, but we explained that they needed to learn the fundamentals first. Overall, they appreciated the chance to explore AI’s strengths and weaknesses. Many found it enlightening to see how AI could help—and where it falls short.
How are you disseminating the work you’re doing to colleagues?
TD: Our faculty recognizes the need to address AI in education, but time and familiarity are barriers. I’m working to create workshops and resources to help faculty at UMR integrate AI into their teaching. There’s a curricular gap here that we’re trying to fill through partnerships and community building. Faculty at UMR are collaborating with AI users at the Mayo Clinic to develop courses on the human side of AI, focusing on its intersection with health care.
Any recommendations for instructors starting to use generative AI in their classes?
TD: My main advice is that whether or not you want to deal with AI, it’s here to stay. We need to reassess our learning objectives and adapt them to include AI. It’s discipline-specific work, but it’s necessary. And it’ll be an ongoing process as technology evolves.
Is there anything else you’d like to share about your journey with generative AI?
TD: Just that I’m grateful for the opportunity to explore this through the fellowship. It’s been a focused and rewarding experience, and it’s great to connect with others tackling similar challenges.
Like what you read? Subscribe to the Extra Points Google Group for an email notification when the next blog post goes live.