The Erosion of LLM Signatures: Can We Still Distinguish Human and LLM-Generated Scientific Ideas After Iterative Paraphrasing?

TL;DR

This paper investigates whether state-of-the-art models can still distinguish between human and LLM-generated scientific ideas after multiple paraphrasing stages, revealing significant challenges and potential improvements with contextual information.

Contribution

It systematically evaluates the decline in detection accuracy of LLM signatures after iterative paraphrasing and explores how contextual information can mitigate this erosion.

Findings

Detection performance drops by 25.4% after five paraphrasing stages.

Incorporating research problem context improves detection accuracy by up to 2.97%.

Simplified, non-expert paraphrasing significantly reduces LLM signature detectability.

Abstract

With the increasing reliance on LLMs as research agents, distinguishing between LLM and human-generated ideas has become crucial for understanding the cognitive nuances of LLMs' research capabilities. While detecting LLM-generated text has been extensively studied, distinguishing human vs LLM-generated scientific idea remains an unexplored area. In this work, we systematically evaluate the ability of state-of-the-art (SOTA) machine learning models to differentiate between human and LLM-generated ideas, particularly after successive paraphrasing stages. Our findings highlight the challenges SOTA models face in source attribution, with detection performance declining by an average of 25.4\% after five consecutive paraphrasing stages. Additionally, we demonstrate that incorporating the research problem as contextual information improves detection performance by up to 2.97%. Notably, our analysis reveals that detection algorithms struggle significantly when ideas are paraphrased into a simplified, non-expert style, contributing the most to the erosion of distinguishable LLM signatures.