A Comparative Analysis of Institutional and Course Generative AI Policies within Higher Education: Implications for Instruction in Computing Education

As generative AI (GenAI) tools like ChatGPT have become ubiquitous, higher education institutions (HEIs) have scrambled to provide policy guidance. This is particularly critical in computer science (CS) education, where tools capable of writing and debugging code directly challenge traditional assessment methods. This study addresses the disconnect between high-level institutional policies and actual classroom implementation by conducting a comparative analysis of institutional guidelines and CS course syllabi from the same set of U.S. research-intensive (R1) universities.

The researchers found a notable divergence between institutional and course-level approaches. At the institutional level, guidance is generally "pro-use," with many universities encouraging faculty to explore GenAI for lesson planning and classroom activities. In contrast, CS instructors are much more cautious. While 92% of the analyzed syllabi provided explicit rules regarding GenAI, 50% of these courses outright prohibited its use.

Furthermore, the study highlights that institutional policies tend to focus on broad ethical concerns—such as data privacy, equity, and academic integrity—whereas course-level policies focus on the "micro" level of instruction. This includes specific requirements for citation, documentation of AI-assisted work, and the use of detection tools. Interestingly, CS instructors frequently anthropomorphize these tools, often referring to them as "assistants" or "partners" in their syllabus language, a trend not mirrored in formal institutional policy documents.

This research reveals that despite institutional encouragement, the "on-the-ground" reality of computing education remains characterized by hesitation and restriction. The authors argue that this inconsistency creates confusion for students and highlights a gap in how faculty are supported in translating broad institutional goals into effective, discipline-specific pedagogical practices. The study concludes that universities must move beyond generic policy statements and provide more concrete, instructor-centered frameworks that help educators integrate AI in ways that support, rather than undermine, the development of essential computational thinking skills.