Stepwise Verification and Remediation of Student Reasoning Errors with Large Language Model Tutors

TL;DR

This paper introduces a stepwise verification approach for LLM-based math tutors that detects student errors more accurately, leading to more targeted and reliable feedback, thereby enhancing personalized education.

Contribution

It presents a novel error verification method grounded in real student solutions, improving LLM tutor responses by accurately identifying mistakes and reducing hallucinations.

Findings

Verifiers improve error detection accuracy.

Targeted responses are more correct and less hallucinated.

Grounding verification enhances tutor quality.

Abstract

Large language models (LLMs) present an opportunity to scale high-quality personalized education to all. A promising approach towards this means is to build dialog tutoring models that scaffold students' problem-solving. However, even though existing LLMs perform well in solving reasoning questions, they struggle to precisely detect student's errors and tailor their feedback to these errors. Inspired by real-world teaching practice where teachers identify student errors and customize their response based on them, we focus on verifying student solutions and show how grounding to such verification improves the overall quality of tutor response generation. We collect a dataset of 1K stepwise math reasoning chains with the first error step annotated by teachers. We show empirically that finding the mistake in a student solution is challenging for current models. We propose and evaluate several verifiers for detecting these errors. Using both automatic and human evaluation we show that the student solution verifiers steer the generation model towards highly targeted responses to student errors which are more often correct with less hallucinations compared to existing baselines.