中国科技期刊研究 ›› 2023, Vol. 34 ›› Issue (4): 446-453. doi: 10.11946/cjstp.202303200186

• 数字出版 • 上一篇    下一篇

ChatGPT探析:AI大型语言模型下学术出版的机遇与挑战

张重毅()(), 牛欣悦, 孙君艳, 祁丽娟, 方梅   

  1. 中国科学院软件研究所期刊联合编辑部,北京市海淀区中关村南四街4号 100190
  • 收稿日期:2023-03-20 修回日期:2023-03-29 出版日期:2023-04-15 发布日期:2023-05-22
  • 作者简介:

    张重毅(ORCID:0000-0001-8511-1671),博士,副编审,E-mail:;

    牛欣悦,硕士,编辑;

    孙君艳,硕士,编辑;

    祁丽娟,博士,副编审;

    方梅,硕士,编审。

    作者贡献声明: 张重毅:设计论文基本框架,撰写论文; 牛欣悦:完成测试,撰写相关测试结论; 孙君艳:调研文献,对比中英文内容,参与论文撰写; 祁丽娟:整理最新研究进展数据; 方梅:把控整体立意及框架结构。

ChatGPT: Opportunities and challenges of large language models for academic publishing

ZHANG Zhongyi()(), NIU Xinyue, SUN Junyan, QI Lijuan, FANG Mei   

  1. Editorial Department of Joint Journal of Institute of Software, Chinese Academy of Sciences, 4 South Fourth Street, Zhongguancun, Haidian District, Beijing 100190, China
  • Received:2023-03-20 Revised:2023-03-29 Online:2023-04-15 Published:2023-05-22

摘要:

【目的】 聚焦大型语言模型(Large Language Models,LLM)的发展表现,针对学术出版行业的特点,分析LLM带来的发展机遇及挑战。【方法】 简要介绍LLM的一般原理,以ChatGPT为案例,测试LLM在处理定义性及开放性问答、总结全文并生成摘要、中英文翻译等方面的表现,进而分析LLM的应用功能及优缺点,并结合学术出版工作内容,分析其对学术出版的影响。【结果】 LLM的快速发展给学术出版行业带来发展机遇,包括稿件内容改进、审稿流程优化、日常工作增速、知识服务转型、媒体融合提质、多语言传播途径扩展等。同时,也存在AI工具使用程度不易把握、隐性学术不端行为更不易识别、科研伦理问题更加不易发现、现行同行评议制度权威性不易保证、著作权及研究成果分配不易处理等方面的挑战。【结论】 学术出版行业要清晰认识自身的核心优势和定位,借AI发展顺势而为,在技术革新的浪潮中更好更快地发展。

关键词: 人工智能, 大型语言模型, ChatGPT, 学术出版, 机遇及挑战

Abstract:

[Purposes] This paper aims to analyze the opportunities and challenges brought by large language models (LLM) to the development of the academic publishing industry with a focus on the development performance of LLM. [Methods] Firstly, the general principles of LLM were briefly introduced. With ChatGPT as an example, the performance of LLM was tested in dealing with such tasks as answering definitive and open questions, summarizing the full texts and generating abstracts, and translating Chinese/English texts. Then, the functions and advantages/disadvantages of LLM were discussed, and the impact of LLM on academic publishing was analyzed with consideration to the academic publishing process. [Findings] The rapid development of LLM has brought opportunities to the academic publishing industry, which include improving the manuscript content, optimizing the peer-review process, accelerating the routine work, transforming knowledge services, improving the quality of media convergence, and expanding multilingual communication ways. However, challenges may come along, such as the difficulties in controlling the usage of artificial intelligence (AI) tools, identifying the hidden academic misconducts, finding problems of research ethics, ensuring the authority of the current peer-review system, and handling the allocation of copyrights and research achievements. [Conclusions] The practitioners of the academic publishing industry should have a clear understanding of its core strengths and positioning for better and faster development in the wave of technical innovation with the aid of AI.

Key words: Artificial intelligence, Large language model, ChatGPT, Academic publishing, Opportunity and challenge