In less than four years after it made its public debut, generative artificial intelligence has achieved near-universal levels of use on college campuses.
Students are using AI to brainstorm ideas, get tutoring help, study for exams, and check over their writing. Institutional attempts to regulate AI use — which are most focused on preventing cheating — have been slow to keep up. One recent survey found that only a quarter of American colleges and universities have formal policies governing AI use. Meanwhile, the tools that faculty and institutions have deployed to root out violations of academic integrity rules have, charitably, turned out to be unreliable and inaccurate.
By now, institutional leaders should realize that further attempts to prevent students from using AI will prove futile. Instead, institutions should look for a way to help their students work with AI — not against it. Students don’t want their colleges to act as AI cops. Students need an AI coach.
As a faculty member, department chair, dean, and now provost at the same university for the past 35 years, I’ve seen instances where the trust between professors and their students erode. Back when I began teaching, taking shortcuts on assignments meant students had to venture into the university library, do a little research, and copy a passage from a relevant article or book. More often than not, the professor was probably familiar with the source material. In some instances, the professor was its author. Whatever the case, it made cheating difficult.
The internet made cheating much easier because students could simply copy and paste text from one digital document to another. It became much more challenging for faculty to detect plagiarized work because their students suddenly had nearly all the reference materials in the world at their fingertips.
Like many other institutions, the University of Nevada, Reno took action to address this new style of cheating, adopting the plagiarism-detection software Turnitin for faculty use. Although Turnitin wasn’t perfect, it mostly did what we wanted it to do — flag potential violations of our academic integrity policy, alleviate some of the pressure on faculty, and encourage students to do their own work or risk receiving a lower grade.
Three years ago, Turnitin introduced its own AI detection tool just as our university’s leaders and faculty were growing increasingly concerned about how generative AI tools like ChatGPT were quickly affecting how students were learning and performing in the classroom. But as everyone in higher education quickly discovered, AI detection tools are notoriously unreliable. High error rates and bias against non-native English speakers could lead faculty to falsely accuse students of academic wrongdoing.
This untrustworthy yardstick put faculty in a tight spot. If they don’t use detection software, they risk students getting away with not doing their own work. And if they do use it, it can raise students’ stress levels and ratchet up their concern that they’ll be wrongly accused of using AI.
University of Nevada students certainly felt stress and concern as well. Being charged with a violation of the university’s academic standards policy can lead to grade reductions, which in turn can jeopardize scholarships and a student’s academic standing. Almost overnight, AI had morphed from something that would revolutionize teaching and learning to a technology whose use generated fear and mistrust on campus and pitted students and faculty against one another. This was not the campus culture we wanted to foster.
So we changed approaches. As part of our transformational PACK AI initiative that weaves AI into our research, teaching and learning across the institution, we took a hard look at our institutional policies to ensure that they met the moment. We wanted to enable students to use AI effectively and responsibly to help them achieve more in college and their careers. We wanted to give faculty the practical strategies, support, and confidence they needed to ensure their students are actively engaging with course material instead of passing the buck to an algorithm. To prioritize student digital privacy, improve student engagement, and reduce the burden on faculty, we reconsidered the AI tools and programs we permitted on campus.
One big decision we made was to get rid of our AI detection software. Rather than using a platform that polices students’ use of AI, we wanted a tool that supported it. That led to a collaboration with Packback, whose platform coaches students throughout the writing process instead of trying to catch them in the act of cheating.
Because AI isn’t going away, our job as educators should be to teach students to use it thoughtfully, not fearfully. UNR faculty have been enthusiastic about moving from cops to coaches. That has created space for more honest and more meaningful engagement with the curriculum. When faculty are allowed to teach and are supported by cutting-edge technology that can coach, students ultimately will be better off.
Read Full Article »