Abstract
The rapid expansion of generative artificial intelligence has changed how students seek help, complete programming work, and demonstrate learning in computer science courses. In the past, computer science students were expected to rely on course faculty, classmates, textbooks, and tutors for support. Now, generative AI tools such as ChatGPT, Claude, and GitHub Copilot can debug code, explain concepts, and even generate assignment solutions. This has created a challenge in computer science because resources for help have not only become abundant, but the meaning of “help” itself has also changed. Instructors have been faced with addressing this shift in an effort to ensure that students are still learning, despite the advanced forms of assistance they can receive outside of the classroom. My technical project focuses on improving a course management system for instructors, teaching assistants, and students in order to facilitate course engagement. My STS research paper explores how syllabi for introductory programming courses at the University of Virginia have changed in response to the prevalence of generative AI, specifically in the ways they define acceptable assistance and how responsibilities between teachers and students have shifted.
My technical report describes advancements to the AI-Smart Classroom Initiative system, or ASCI, a course management system designed to integrate multiple tools, such as an office hours queue, course-specific chatbots, administrative tools, and course gamification features. The system was designed to address issues in large classrooms, such as long office hours wait times, lack of engagement with office hours, and the use of external generative AI. My capstone team improved ASCI by improving content uploads to the chatbot, adding administrative tools, creating a Discord activity monitor, finishing the quest system, and adding a course archival feature that allows instructors to deactivate courses when a new semester begins. The biggest advancement was the quest system, through which instructors can create extra credit opportunities that incentivize students to increase their engagement with the course. The evaluation of these improvements was conducted through stakeholder review, in which we asked teaching assistants, students, and the original designers of the system to test and critique our improvements. Although the feedback was generally positive, we were unable to conduct a more accurate assessment of the impact of these improvements, which would require testing the system in actual courses, monitoring grades over an extended period of time, and collecting a larger dataset. Despite these limitations, our improvements to the ASCI system suggest that a course-support platform can reduce administrative tasks for instructors in high-enrollment courses and encourage students to engage with a course more effectively.
My STS research paper explores an issue more directly focused on computer science: how generative AI has shifted responsibility, norms, and expectations in introductory computer science courses, as reflected in UVA course syllabi. I researched this question by examining syllabi because they are official course documents that serve as the primary means through which instructors define expectations, responsibilities, collaboration, and acceptable assistance for their students. My analysis compared older UVA introductory computer science syllabi from before the prevalence of generative AI with syllabi from the AI era, specifically from 2024 and 2025. The main finding from this analysis was that expectations for academic integrity and acceptable forms of course support were extended, not replaced. Syllabi used prior to the prevalence of generative AI emphasized authenticity of work, responsible collaboration, and students’ understanding of the work they submitted as their own. AI-era syllabi still stated these same concerns, but they also introduced new expectations, such as citation of sources, disclosure, and conditions on the use of resources. These new expectations reveal a shift from simply expecting students to submit their own work to also expecting them to disclose the assistance they may have received. The difficulty of regulating generative AI usage outside of the course creates the need for students to bear the responsibility of using generative AI transparently. My STS paper reveals that introductory programming courses must assess programming assignments based on the development process, rather than just the final product.
The technical project and STS research paper show that generative AI is more than a technical issue in education. It presents a challenge to the current sociotechnical system that defines education. The ASCI project advancements explore one possible technical effort to create a system that provides students with structure when using innovative tools to supplement their learning. The STS research paper examines how learning in computer science is disrupted when structured systems for regulating generative AI usage are absent. Together, these projects reveal that creating and modifying systemic structures to promote transparent and accountable processes is the approach most conducive to productive education from both social and technical standpoints.
