Whodunit: Classifying Code as Human Authored or GPT-4 Generated -- A case study on CodeChef problems

TL;DR

This paper addresses the problem of distinguishing GPT-4 generated from human-authored Python code, focusing on educational integrity. It proposes a code-stylometry approach using a CodeChef-derived dataset and an XGBoost classifier, augmented with SHAP explanations for interpretability. The main finding is a robust $F1$-score and $AUC$-ROC of about 0.91, with the average line length emerging as the strongest discriminative feature; even when excluding gameable features, performance remains high. The study demonstrates the viability of a scalable, interpretable detector for AI-generated code, while acknowledging limitations in generalizability across platforms, languages, and AI models and outlining directions for broader evaluation and robustness enhancements.

Abstract

Artificial intelligence (AI) assistants such as GitHub Copilot and ChatGPT, built on large language models like GPT-4, are revolutionizing how programming tasks are performed, raising questions about whether code is authored by generative AI models. Such questions are of particular interest to educators, who worry that these tools enable a new form of academic dishonesty, in which students submit AI generated code as their own work. Our research explores the viability of using code stylometry and machine learning to distinguish between GPT-4 generated and human-authored code. Our dataset comprises human-authored solutions from CodeChef and AI-authored solutions generated by GPT-4. Our classifier outperforms baselines, with an F1-score and AUC-ROC score of 0.91. A variant of our classifier that excludes gameable features (e.g., empty lines, whitespace) still performs well with an F1-score and AUC-ROC score of 0.89. We also evaluated our classifier with respect to the difficulty of the programming problem and found that there was almost no difference between easier and intermediate problems, and the classifier performed only slightly worse on harder problems. Our study shows that code stylometry is a promising approach for distinguishing between GPT-4 generated code and human-authored code.

Figures (8)

• Figure 1: Solutions to a palindrome problem from CodeChef
• Figure 2: An overview of our approach in detecting GPT-4 generated code
• Figure 3: Prompt used for generating 2 AI code solutions
• Figure 4: SHAP feature importance of our approach
• Figure 5: SHAP feature importance of non-gameable features