"Everyone's using it, but no one is allowed to talk about it": College Students' Experiences Navigating the Higher Education Environment in a Generative AI World

A University of Washington research team led by Yue Fu has published a revealing study on how college students navigate generative AI in academic settings. Based on semi-structured interviews with 23 students, the paper documents a significant disconnect between institutional policies and actual student behavior. The research identifies several key drivers: environmental pressures like grading cycles and deadlines force students to use AI tools for efficiency, even when they personally believe it compromises their learning outcomes. Social dynamics within peer micro-communities establish de-facto norms that often contradict official guidelines, creating parallel systems of AI ethics. The study coins the term 'AI shame' to describe the widespread phenomenon where students feel compelled to hide their AI usage due to perceived stigma, pushing these practices underground. Current institutional AI policies are described by participants as 'generic, inconsistent, and confusing,' resulting in what researchers term 'routine noncompliance.' Students reported developing personal value-based frameworks for AI use, but environmental pressures consistently created gaps between their intentions and actual behaviors. The findings suggest AI usage in academia is a 'situated practice' heavily influenced by context rather than abstract policy. The researchers conclude with implications for institutions, instructors, and tool designers, emphasizing the need for more nuanced approaches that acknowledge rather than deny the reality of widespread AI adoption.