Abstract
Our study assesses the knowledge and perceptions that postsecondary writing instructors have of generative AI programs such as ChatGPT and tests instructors’ ability to distinguish between essays written entirely by students and essays generated by ChatGPT. We tested and interviewed twenty experienced postsecondary instructors currently teaching writing. Participants graded four essays and attempted to identify which essays were AI-generated. We found that writing instructors have a moderate level of confidence in their ability to distinguish between student and AI-generated writing but a low level of accuracy—only 35% of instructors could correctly identify the authorship of all four essays. AI-generated essays scored higher than essays written by students, especially in spelling, grammar, and organization, while they scored lower in argumentation and evidence. We suggest that instructors will need to find ways to encourage students to work independently while learning to use AI as a writing support. In our conclusion, we discuss pedagogical solutions that allow the use of AI and propose that these solutions can complement administrative ones.