Student and Faculty Perceptions of Generative Artificial Intelligence in Student Writing

Abstract

Background Psychology instructors frequently assign writing-to-learn exercises that include personal reflection. Generative Artificial Intelligence (GenAI) can write text that passes for humans in other domains. Objective Do students and faculty rate a reflection written by GenAI differently than reflections written by students? Do students and faculty agree about the appropriateness of using GenAI for college-level writing? Method Eighty-three students and 82 faculty read four reflections (three written by undergraduate students and one by GenAI). After rating the quality of each, they chose which one they thought was AI-generated. Participants then rated the ethicality of nine potential ways to use GenAI in college-level writing and the potential of each to compromise learning. Results Participants rated the AI-generated reflection similarly to the student-generated reflections and failed to reliably detect AI-generated writing. Faculty and students agreed that using GenAI to produce the final text for a student likely compromises learning more than using it to generate ideas. Conclusion AI-generated reflections blend in with student-written reflections, and students and faculty agree about the potential detriments to learning. Teaching Implications GenAI can be hard to detect in the psychology classroom. Rather than implementing one-size-fits-all policies, instructors might focus classroom conversations on how GenAI could compromise learning.