Can AI replace teachers in grading and feedback provision?
Research by Arnaud Chevalier, Jakub Orzech and Petar Stankov using a randomised controlled trial (RCT) show that undergraduate students receiving feedback from ChatGPT 4 perform as well in subsequent assessments than their peers in the control group who always received human feedback. Separately, ChatGPT grades and ranks students significantly differently from a teacher. Worryingly, the same assessment receives largely different grades when marked repeatedly by ChatGPT. Overall, while the newest LLM helps learning as well as a teacher, its ability to grade student work is still inferior.
Read the full paper at:
https://www.iza.org/publications/dp/17511/man-vs-machine-can-ai-grade-and-give-feedback-like-a-human