By Paul A. Djupe, Director of Data for Political Research

Last year AI felt like it would hit academia like the Chicxulub impactor – the asteroid that killed off the dinosaurs. To the Eeyores among us, ChatGPT spelled the end of the college essay, perhaps the end of academia itself. If students can generate viable enough papers in 10 seconds of prompt writing, then what are we doing here? Of course, AI large language models are not the first significant disruption to how colleges and universities function; it wasn’t that long ago that the same sorts of arguments were trotted out about the role of the internet. But this is new and no one is sure exactly where we’re headed and what we should do about it or with it. One thing we at 127 will continue to do is track student use of AI as well as how they think about it. Toward that end, here’s our first snapshot using October 2023 survey data of about 500 students.

We asked about a range of possible uses prefaced with the disclaimer that admissions here are not violations of academic integrity because they are not tied to any course or assignment. It’s clear that a majority has used AI for something course related, but there’s diversity in what that is. Mostly students are using AI, like ChatGPT, Bard, or Bing, for background research and summarizing readings. Only a small minority admit to using AI to write a paper. Clearly some are using AI for more than one task, but they are not common. Just 1 percent admits to all uses and 40 percent have used AI for 1-2 tasks. So, usage on campus is fairly broad, and certainly not deep.

Of course we’re interested in who is using AI and a couple of groups stand out. First years are less likely to use AI, as are women. It’s interesting that GPA is not linked to AI use, perhaps because AI can be both an assistant helping with efficiency and a savior bailing out the overwhelmed student. But one group stands out way above the rest in their AI use – international students.

Seen in the graph below, international students admit to AI use at double or triple the rates that domestic students do. They admit to using AI to write a paper (18%) at 3x the rate of others (6%). Majorities admit to using AI for background research and reading summaries – 2x the rate of other students. This pattern makes sense if cheating is linked to being overwhelmed, which I assume is the deal here – writing at a high level in a second language would be very difficult.

I also looked to see if first-generation students used it more or less. I didn’t really know what to expect here – first-gen students might be a bit more overwhelmed since they may not have the same socialization into what to expect in college, but maybe they are less aware of “study aids” because of that presumed lack of socialization. So it’s interesting to note that 15 percent more domestic first-gen students admit to not using AI. In each category of task, fewer first-gen students said they used AI. Perhaps we might reframe our thinking about AI in light of these results – it could be seen as an equity issue that the college might help remedy.

I want to close by showing a couple of attitudes about AI that students have, which may vary by major. Overall, 40 percent of students agree that AI undermines the value of a college education, but the response differs considerably by major. As I expected, the arts and humanities majors lead the way with majorities (or nearly so) in agreement. Science and interdisciplinary students are the least likely to agree, perhaps because they have already seen useful applications. What is striking about these results is the amount of ambivalence – near majorities of interdisciplinary students are on the fence. And of course large numbers of the undecided about their major are also undecided about AI.

There are many other questions that we asked, but I’d like to close this piece by asking if students want more opportunities to use AI in their coursework. Who is surprised that the undecided are ALL IN for AI use. But of those committed to a major, there are no majorities that want more AI use. The closest are the science majors – 47 percent of whom want more AI. Paralleling the last figure, arts and humanities majors are the least receptive to AI use opportunities. But, again, there are large numbers who are on the fence. That’s telling. I think we just don’t know what those opportunities would entail, so it’s hard to commit to it at this early stage in AI’s rollout.

It’s early. We’ve only had AI in our lives for a few years at most and I think it’s clear that its use will only grow in scope in the years to come. While eventually we’ll have to offer courses in how to defeat SkyNet, we’re not quite ready to have majors in Sarah Connorism. For now, there’s a lot of student ambivalence about AI that will settle into firmer attitudes once we have some experience using it balanced by its educational value. We’ll ask most of these questions again this spring to get another read on it to see how quickly attitudes and usage are changing.

Paul A. Djupe is a currently laid-up local cyclist who runs the Data for Political Research minor. He started onetwentyseven.blog a few years ago in a bid to subsidize collective action and spread accurate knowledge about campus and what goes on there. He also writes about religion and politics in the US.