PnB-M10 - Promote Responsible AI Use through Peer Review Interventions 

Information

In this poster, we document two peer review interventions aimed to promote self-regulated learning and responsible AI use in student work. The first intervention is an assignment-activity design with generative AI output critically examined in peer review. The design was implemented in two sections of the Introductory to Data Science course offered to undergraduate students at Michigan State University (MSU) in Fall '25. Each section enrolls around 60 students from statistics or data science majors. Students were asked to use generative AI to complete a take-home assignment consisting of short answer questions in statistics, such as describing a scatterplot. During the lecture discussion, students peer review each other's AI output, grade them based on the rubric provided by the instructor, and share their findings. Student verbal feedback and voluntary surveys show that through this activity, students begin to a) take a critical perspective on generative AI, b) recognize the ambiguity and redundancy from generative AI responses, and c) have a better understanding of the statistics questions. The second intervention implemented peer-review in a cohort-based graduate data science course offered to Master of Science in Data Science students at MSU during their second semester. Roughly ~90 total students in Spring '25 and '26 were asked to review other groups' final project drafts before submission. Compared to assignments turned in during Fall '25 & '26, work grew in quality after being peer-reviewed. Notably, work was more concise, much less noticeably generated from AI, and better formatted. Additionally, the instructor spent less time providing feedback and grading resubmissions. Student feedback from the 25/26 AY cohort will be gathered and presented at the conference.