Limits of an AI program for solving college math problems

TL;DR

This paper critically examines a neural network-based system claiming to solve university math problems, revealing that it primarily relies on symbolic algebra software and test data rather than genuine problem-solving capabilities.

Contribution

The paper provides a detailed critique of prior claims, highlighting the system's reliance on symbolic computation, limited problem formats, and potential data leakage, challenging its purported human-level performance.

Findings

System uses Sympy for solving, not neural networks

Limited to specific problem formats, excluding many types

Potential use of test answers to guide solutions

Abstract

Drori et al. (2022) report that "A neural network solves, explains, and generates university math problems by program synthesis and few-shot learning at human level ... [It] automatically answers 81\% of university-level mathematics problems." The system they describe is indeed impressive; however, the above description is very much overstated. The work of solving the problems is done, not by a neural network, but by the symbolic algebra package Sympy. Problems of various formats are excluded from consideration. The so-called "explanations" are just rewordings of lines of code. Answers are marked as correct that are not in the form specified in the problem. Most seriously, it seems that in many cases the system uses the correct answer given in the test corpus to guide its path to solving the problem.