Classifying Permissiveness in Generative AI Syllabus Policies: A Corpus-Based Typology Development

• In Event: AERA Poster Session Friday 7:45 am
  In Poster Session: ATL Poster Session 2: AI Integration and Outcomes in Postsecondary Settings

Abstract

This study explores the variability in generative AI (GenAI) sample syllabus policy statements by developing a systematic typology of permissiveness (Permissive, Flexible Permissive, Flexible Restrictive, Restrictive). An institutional corpus of sample policy statements was analyzed using a human-in-the-loop triangulation with ChatGPT 4, Claude Sonnet 4, and Mistral. The analysis identified four distinct categories based on linguistic and rhetorical patterns. The resulting typology standardizes terminology across institutions, enables cross-institutional comparison, and uncovers how institutional labels often misalign with policy language.

Authors

• Amanda Taylor, Boise State University

• Brett E. Shelton, Boise State University