When I am in a headspace to think rationally about generative AI, I often think about Melvin Kranzberg’s first law of technology: “Technology is neither good nor bad; nor is it neutral” (1986: 545). Most technology simultaneously produces benefits and harms. The outcomes are context dependent and unintended consequences are practically guaranteed.
As a risk adverse person, I was fairly hesitant to incorporate generative AI in my classes. I worried it would stop my students from thinking for themselves, flatten and overly standardize their writing, allow them to turn in polished work they didn’t really understand, and stifle class conversation. But it was clear generative AI was in my classroom whether I gave it permission to enter or not. I’ve never had so many students “delving” into a topic, “underscoring” important points in their writing, assertively stating the overarching themes a novel when they had only read two chapters, and boldly proclaiming “it’s not x, it’s y” about every topic imaginable. So I thought I might as well see if I could channel its use in some way by incorporating it into an assignment.
I chose to allow students to incorporate generative AI into a group presentation assignment for an introductory Criminology class. Groups can choose their topic, but the presentation must do a few things:
- Give an overview of the criminological issue, debate, or puzzle.
- Review and apply something we’ve talked about in class and discuss how class material helps make sense of the group’s topic.
- Involve interactive elements (group discussion, an activity, something so it’s not just a group lecture).
- Convey why what the group is discussing is interesting.
- Come together as a complete package. The presentation should feel like the elements are connected to one another and ultimately answer some version of a question like “so what?” or “why should we care about this topic?”
The topics students choose are almost always interesting. A sampling from the past two semesters include:
- An application of Agnew’s general strain theory to the salon bandit case (a series of robberies that took place at gas stations, a Bath and Body Works, and tanning salons).
- How social media use impacted the investigation into the Idaho Four case.
- An application of the fraud triangle to romance scams (and the Dirty John case in particular).
- CTE (Chronic Traumatic Encephalopathy, head trauma caused by repeated impact) and violent crime amongst current and former NFL players.
Students had the option to use generative AI to brainstorm, help them refine their topics, apply class concepts, generate a presentation outline to streamline the flow of the presentation, and draft discussion questions. Some groups chose not to use AI, but most used it in at least some capacity.
I’ve taught this class twice and used a version of the presentation assignment each time. The first time, there was no option to use AI (although I’m sure some students used it anyway). The second time students could use AI. Overall, the presentations were well done and interesting each time (perhaps unsurprisingly, students like talking about crime and found their topics intriguing, which led to good presentations). I don’t have enough data to definitely say how AI use impacted their presentations, but I observed three general trends.
In the positive column, generative AI use seemed to help students identify and clearly convey a guiding topic. Many students can do this on their own, but if students struggle with something in this presentation assignment, it’s pulling all of their tasks and information into a cohesive package that they can convey to an audience. AI use seemed to make some groups more aware of the need to think about how all of their elements of their presentation fit together. They were more likely to articulate a guiding question or focus and come back to this guiding question throughout the presentation.
In the negative column, if students struggled with anything, it was something I was unprepared for: too much information. Some of the groups that used AI had presentations that were overly dense with information. When you have easy access to so much information, you want to throw it all in and see what happens. Unfortunately, what often happens is you and the audience get bogged down in too many details.
In the puzzling column, generative AI use seemed to give these presentations a flatter quality. Similar to the discussions about generative AI use in writing, the presentations that employed the most generative AI were solid, technically proficient, mechanically thorough presentations. But they had less of that sparkly energy, that unpredictability, that profound messy humanness that comes from an undergraduate student group presentation. It’s not something I knew I would miss until it was gone.
Many of my courses have some form of group presentation element. I’ll allow students to use generative AI in the future, but I’ll have to think about how to prompt students to cull the information more. I’ll also have to figure out how to incentivize students to take more chances, be more human, and occasionally resist generative AI’s siren call of technically competent, but slightly bland, mechanical proficiency. When I figure out how to do that, I’ll let you know.
