Saved in:
Bibliographic Details
Main Authors: Patel, Manooshree, Bhattacharyya, Rayna, Lu, Thomas, Mehta, Arnav, Voss, Niels, Norouzi, Narges, Ranade, Gireeja
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.08321
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908863807094784
author Patel, Manooshree
Bhattacharyya, Rayna
Lu, Thomas
Mehta, Arnav
Voss, Niels
Norouzi, Narges
Ranade, Gireeja
author_facet Patel, Manooshree
Bhattacharyya, Rayna
Lu, Thomas
Mehta, Arnav
Voss, Niels
Norouzi, Narges
Ranade, Gireeja
contents This paper considers the development of an AI-based provably-correct mathematical proof tutor. While Large Language Models (LLMs) allow seamless communication in natural language, they are error prone. Theorem provers such as Lean allow for provable-correctness, but these are hard for students to learn. We present a proof-of-concept system (LeanTutor) by combining the complementary strengths of LLMs and theorem provers. LeanTutor is composed of three modules: (i) an autoformalizer/proof-checker, (ii) a next-step generator, and (iii) a natural language feedback generator. To evaluate the system, we introduce PeanoBench, a dataset of 371 Peano Arithmetic proofs in human-written natural language and formal language, derived from the Natural Numbers Game.
format Preprint
id arxiv_https___arxiv_org_abs_2506_08321
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LeanTutor: Towards a Verified AI Mathematical Proof Tutor
Patel, Manooshree
Bhattacharyya, Rayna
Lu, Thomas
Mehta, Arnav
Voss, Niels
Norouzi, Narges
Ranade, Gireeja
Artificial Intelligence
Human-Computer Interaction
Logic in Computer Science
This paper considers the development of an AI-based provably-correct mathematical proof tutor. While Large Language Models (LLMs) allow seamless communication in natural language, they are error prone. Theorem provers such as Lean allow for provable-correctness, but these are hard for students to learn. We present a proof-of-concept system (LeanTutor) by combining the complementary strengths of LLMs and theorem provers. LeanTutor is composed of three modules: (i) an autoformalizer/proof-checker, (ii) a next-step generator, and (iii) a natural language feedback generator. To evaluate the system, we introduce PeanoBench, a dataset of 371 Peano Arithmetic proofs in human-written natural language and formal language, derived from the Natural Numbers Game.
title LeanTutor: Towards a Verified AI Mathematical Proof Tutor
topic Artificial Intelligence
Human-Computer Interaction
Logic in Computer Science
url https://arxiv.org/abs/2506.08321