Can Large Language Models Detect Methodological Flaws? Evidence from Gesture Recognition for UAV-Based Rescue Operation Based on Deep Learning

TL;DR

This study demonstrates that large language models can identify methodological flaws like data leakage in machine learning research papers by analyzing published artifacts without prior context.

Contribution

It shows that LLMs can independently detect common methodological issues in scientific papers, aiding reproducibility and scientific auditing.

Findings

All six LLMs identified data leakage in the case study.

Models attributed high accuracy to non-independent data splits.

LLMs showed potential as tools for scientific artifact analysis.

Abstract

Reliable evaluation is essential in machine learning research, yet methodological flaws-particularly data leakage-continue to undermine the validity of reported results. In this work, we investigate whether large language models (LLMs) can act as independent analytical agents capable of identifying such issues in published studies. As a case study, we analyze a gesture-recognition paper reporting near-perfect accuracy on a small, human-centered dataset. We first show that the evaluation protocol is consistent with subject-level data leakage due to non-independent training and test splits. We then assess whether this flaw can be detected independently by six state-of-the-art LLMs, each analyzing the original paper without prior context using an identical prompt. All models consistently identify the evaluation as flawed and attribute the reported performance to non-independent data partitioning, supported by indicators such as overlapping learning curves, minimal generalization gap, and near-perfect classification results. These findings suggest that LLMs can detect common methodological issues based solely on published artifacts. While not definitive, their consistent agreement highlights their potential as complementary tools for improving reproducibility and supporting scientific auditing.