logo
科言猫学术研究的AI总结
language
手机端手机端
  • Phone Feature微信小程序
简体中文简体中文
notice消息通知
博客博客
delete
arrow
返回

Quantifying large language model usage in scientific papers

delete2025-08-04
delete0
PRE
AI
W
Weixin Liang *
Y
Yaohui Zhang
Z
Zhengxuan Wu
H
Haley Lepp
W
Wenlong Ji
X
Xuandong Zhao
H
Hancheng Cao
刘
刘胜 (Sheng Liu)
何思雨 封面图
何思雨 (Siyu He)
Z
Zhi Huang
D
Diyi Yang
C
Christopher Potts
C
Christopher D. Manning
J
James Zou *
DOI：10.1038/s41562-025-02273-8delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
科学出版是传播研究成果的主要途径。已有关于大型语言模型（LLMs）在学术写作中广泛使用的推测。在此，我们对2020年1月至2024年9月期间arXiv、bioRxiv和Nature出版集团期刊上的1,121,912篇预印本和已发表论文进行了系统性分析，采用基于词频变化的人口水平框架来估计LLM修改内容随时间的流行程度。我们的发现表明，LLM的使用呈稳步增长趋势，其中计算机科学论文的增长规模最大且速度最快（最高达22%）。相比之下，数学论文和Nature出版集团期刊显示出较低的LLM修改证据（最高达9%）。LLM修改估计值在更频繁发布预印本的论文第一作者、研究领域更拥挤的论文以及篇幅较短的论文中更高。我们的发现表明，LLMs正被广泛用于科学写作。Liang等人估计了近期科学论文和预印本中经大型语言模型修改的文本的流行程度，发现其使用广泛（计算机科学论文中高达17.5%）。

期刊

Nature Human Behaviour 封面图
Nature Human Behaviour
IF:
15.9
论文数:
504
被引数:
1.7W

机构

D
department of electrical engineering
学者数:
1.3K
论文数: 671
被引数: 0
D
Department of Computer Science
学者数:
1.7K
论文数: 998
被引数: 8
D
department of biomedical data science
学者数:
30
论文数: 22
被引数: 0
G
Graduate School of Education
学者数:
137
论文数: 104
被引数: 0
D
department of statistics
学者数:
581
论文数: 369
被引数: 4
学者 查看更多机构
引用论文

引用论文

Automatic Detection of Generated Text is Easiest when Humans are Fooled
errProceedings of the 58th Annual Meeting of the Association for Computational Linguistics
IF0
err2020-01-01
err0
errOAAI
errDaphne Ippolito; Daniel Duckworth; Chris Callison-Burch; Douglas Eck
err分享
err收藏
GPT detectors are biased against non-native English writers
errPATTERNS
IF7.4
err2023-07-01
err66
errOAAI
errLiang, Weixin; Yuksekgonul, Mert; Mao, Yining; Wu, Eric; Zou, James
err分享
err收藏
ChatGPT is fun, but not an authorChatGPT很有趣，但不是作者
errScience
IF0
err2023-01-27
err0
errOAAI
errH. Holden Thorp
err分享
err收藏
Testing of detection tools for AI-generated text测试AI生成文本的检测工具
errINTERNATIONAL JOURNAL FOR EDUCATIONAL INTEGRITY
IF6.9
err2023-12-25
err70
errOAAI
errWeber-Wulff, Debora; Anohina-Naumeca, Alla; Bjelobaba, Sonja; Foltynek, Tomas; Guerrero-Dib, Jean; Popoola, Olumide; Sigut, Petr; Waddington, Lorna
err分享
err收藏
S2ORC: The Semantic Scholar Open Research CorpusS2ORC: 语义学者开放研究语料库
errProceedings of the 58th Annual Meeting of the Association for Computational Linguistics
IF0
err2020-01-01
err0
errOAAI
errKyle Lo; Lucy Lu Wang; Mark Neumann; Rodney Kinney; Daniel Weld
err分享
err收藏
What Is a Paraphrase?
errCOMPUTATIONAL LINGUISTICS
IF5.3
err2013-09-01
err117
errOAAI
errBhagat, Rahul; Hovy, Eduard
err分享
err收藏
err
IF0
err
err0
PREAI
err
err分享
err收藏
How Familiarity, Ease of Use, Usefulness, and Trust Influence the Acceptance of Generative Artificial Intelligence (AI)-Assisted Travel Planning
errINTERNATIONAL JOURNAL OF HUMAN-COMPUTER INTERACTION
IF4.9
err2024-11-11
err0
PREAI
errTopsakal, Yunus
err分享
err收藏
Can Large Language Models Provide Useful Feedback on Research Papers? A Large-Scale Empirical Analysis
errNEJM AI
IF0
err2024-07-25
err0
errOAAI
errWeixin Liang; Yuhui Zhang; Hancheng Cao; Binglu Wang; Daisy Yi Ding; Xinyu Yang; Kailas Vodrahalli; Siyu He; Daniel Scott Smith; Yian Yin; Daniel A. McFarland; James Zou
err分享
err收藏
MAGE: Machine-generated Text Detection in the WildMAGE：开放场景下的机器生成文本检测
errProceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
IF0
err2024-01-01
err0
PREAI
errYafu Li; Qintong Li; Leyang Cui; Wei Bi; Zhilin Wang; Longyue Wang; Linyi Yang; Shuming Shi; Yue Zhang
err分享
err收藏
学者 查看更多内容