Experiences with GenAI in a Geology Class

By: Christine Clark

By Summer 2023, I had heard a good amount about GenAI, most of it frightening for an educator. I went to the annual meeting of my professional society that fall, hopeful to find out how others had addressed the myriad of concerns I had. To my surprise, no one was talking about it. I realized that my best option was to just jump in head first, and see what came out of it.

In the fall of 2023, I was teaching the Writing Intensive course in the Geology program, ESSC 466W, Global Tectonics. Throughout the course, we discuss reading scientific, peer-reviewed articles. Students are assigned articles to read which they then have to respond to comprehension questions about. These questions not only assess their understanding of the article, but the methodology the authors used and whether their conclusions are valid based on their observations. I decided to use one of these assignments to have the students assess how well a GenAI could respond to the same questions. After having the students read and respond to a reasonably challenging article (concerning slip lineations in the Sierra Nevada Mountains), the students were then asked to submit a portion of the analysis questions to a GenAI. They were to evaluate how well the GenAI was able to respond to the prompts; we then discussed all the responses in class.

I was pleasantly surprised with how well this assignment worked. The students were able to see for themselves that the GenAI could provide surficial information about the topic, but could not analyze it at deeper levels. These are some of the student responses to the question about how well GenAI did:

• The responses provided by chat GPT generally showed a lack of in-depth understanding

of the topics at hand. Some of the answers were fully incorrect.

• I feel like this AI got some things right, and most of it wrong. That or they are saying the same thing just with different descriptions, I don’t know, I’m not an expert in this.

• If you were struggling with understanding those concepts as you encountered them in the article, the summary that it gave might be nice, although those definitions can be found from more reputable sources with a quick google search with the same amount of effort.

• I’m kind of freaked out that Chat GPT did so well in answering these honestly.

• Obviously ChatGPT is limited in its ability to respond to these specific questions because

it apparently can’t look at outside links.

• While the software gave me a great detailed look at the questions, it didn’t really answer the questions I entered into it.

I took what I did, coupled with a comparable assignment given to Introductory Geology students (ESSC 110), back to the same conference this fall. The number of attendees that stopped by my poster to discuss our assignments was staggering. Overall, it seems like there are large numbers of us out there; instructors who want to tackle GenAI head-on in our classes and to present it as a tool with pros and cons. Using GenAI in this way did not take a significant amount of time away from my class, and left the students with the knowledge that GenAI has limitations. This was an experiment in teaching that worked well.

Looking forward, GenAI will become both “smarter” and more seamless in our everyday lives. Already, Gen AI-crafted responses show up on Google searches. The genie is not going back in the bottle on this one; we need to recognize that this technology is here to stay. As educators, we need to give our students the tools they will need to be successful in their careers. 

This semester, in addition to giving the students the same article review assignment using GenAI, I have also explicitly permitted them to use it to assist in writing their semester-long research paper. There are still constraints; for instance, I have asked them to cite the GenAI as a reference if they’ve used it. I am not sure how this will work, but I am willing to learn as I go and try one new thing. I have learned that it just takes a bit of bravery to be willing to learn by experiment. As instructors, it’s important to think about what the goal of any particular assignment is. What is the learning objective? And can the students reach the learning objective using GenAI? If they can’t, fine, don’t let the students use it for that assignment. But if they can still reach the learning objective using GenAI, then perhaps allow it for that assignment. It’s okay to have different policies for different assignments, as long as you are clear and explain the policy to the students with the “why”.

As we prepare our students for their future careers, it’s imperative that they have the chance to learn just what GenAI does well, and what it doesn’t. I encourage you to just try it…and see what happens.