Can Large Language Models Replicate ITS Feedback on Open-Ended Math Questions?

TL;DR

This study investigates whether large language models can generate effective feedback for open-ended math questions in ITS, finding they replicate training data well but struggle with unseen errors, indicating limited understanding of mathematical mistakes.

Contribution

The paper evaluates LLMs' ability to produce math feedback in ITS, highlighting their strengths in formatting but limitations in understanding diverse student errors.

Findings

LLMs replicate training feedback effectively

Models do not generalize well to unseen errors

LLMs show limited understanding of mathematical mistakes

Abstract

Intelligent Tutoring Systems (ITSs) often contain an automated feedback component, which provides a predefined feedback message to students when they detect a predefined error. To such a feedback component, we often resort to template-based approaches. These approaches require significant effort from human experts to detect a limited number of possible student errors and provide corresponding feedback. This limitation is exemplified in open-ended math questions, where there can be a large number of different incorrect errors. In our work, we examine the capabilities of large language models (LLMs) to generate feedback for open-ended math questions, similar to that of an established ITS that uses a template-based approach. We fine-tune both open-source and proprietary LLMs on real student responses and corresponding ITS-provided feedback. We measure the quality of the generated feedback using text similarity metrics. We find that open-source and proprietary models both show promise in replicating the feedback they see during training, but do not generalize well to previously unseen student errors. These results suggest that despite being able to learn the formatting of feedback, LLMs are not able to fully understand mathematical errors made by students.