By Savvy Barnes (Michigan State University)
Information
Use of generative AI in the classroom seems to fall into two extremes: banned, and allowed per student discretion. In this poster, we suggest an alternative: mandatory generative AI assignments with guided examples. This poster documents the inclusion of mandatory, guided generative AI assignments in the second, required statistics course within the Master of Data Science program at Michigan State University. The program enrolls a cohort of ~45 students from diverse backgrounds with various levels of experience with data science. In both required statistics courses, students were allowed generative AI usage with minimal restrictions. However, the second made generative AI usage mandatory on some assignments. Students were asked to take a voluntary survey comparing assignments between the statistics course and the second, with an emphasis on how the inclusion of mandatory generative AI in the second course changed their approach to answering assignments in general. Additionally, questions were asked about their attitude towards AI and perceived changes from one semester to the next. Early observations show students used generative AI on non-mandatory questions less often, or more intentionally, in the second semester compared to the first. Survey collection is ongoing, and complete results will be available at USCOTS 2025.