PhD Candidate Kendall Lowe and Professor Ezra Keshet recently wrote an article for the Center for Research on Learning and Teaching (CRLT) about a project on which they worked over the 2025 summer. In this project, they investigated how well ChatGPT would fare against assignments in LING 111, the department’s Introduction to Linguistics course. They tested a variety of assignment types from the course and how both naive and sophisticated users of ChatGPT would fare in using it to complete assignments, finding that ChatGPT fared surprisingly well in the course. The article also discusses which types of questions were easiest and hardest to answer well using ChatGPT. Stay tuned for a future article which will discuss potential strategies to mitigate the issues discussed in this article.
Read the full article on CRLT’s Substack here.
