The Risk Review and Countermeasures of Using AIGC in Academic Research
|
Zhang Yanli, master, E-mail: zyl11604@cnki.net; |
|
Guan Yuyan, master; Wan Xue, master; Zhao Jinghao, master. |
|
Wan Xue, master; Zhao Jinghao, master. |
|
Zhao Jinghao, master. |
Received date: 2024-05-12
Online published: 2025-04-28
[Purpose/Significance] This study aims to analyze the potential risks of using AIGC tools during academic research, study the strategies to avoid these risks and provide references and guidance for researchers so that they can use AIGC tools in compliance with rules and laws. [Method/Process] This study sorted out the policies and regulations related to the application of AIGC in academic fields at home and abroad, summarized and analyzed the main risks of using AIGC. Then, combined with the researchers’ academic research process, risk avoidance strategies and feasible suggestions for using AIGC at different stages were proposed. [Result/Conclusion] The risks of using AIGC tools for researchers mainly focus on the security of input content, the reliability of output results, academic ethical risks, and potential legal risks. Researchers should enhance their risk-awareness, understand the policy regulations, know the usage boundaries for different research stages and scenarios very well, and regulate their usage behavior to ensure that AIGC tools play a positive role in academic research.
Key words: AIGC; Generative AI; academic research; risk analysis; countermeasure
Yanli Zhang , Yuyan Guan , Xue Wan , Jinghao Zhao . The Risk Review and Countermeasures of Using AIGC in Academic Research[J]. Knowledge Management Forum, 2024 , 9(4) : 341 -352 . DOI: 10.13266/j.issn.2095-5472.2024.025
表1 学术研究中不应使用AIGC工具的场景Table 1 Scenarios where AIGC should not be used in academic research |
| 阶段 | 不应使用场景 | 注意事项 |
| 数据收集与分析 | 生成、伪造或篡改实验、观测类数据; 拼凑、伪造或篡改反应实验、观测过程或结果的图像、图形和音视频等多媒体内容 | 非真实的数据不仅可能导致错误的分析和结论,存在内容可靠性风险,还违反科研诚信的基本原则,会给作者带来学术声誉的损害、论文被撤稿的问题,甚至面临法律责任 |
| 研究设计与研究开展 | 生成研究假设; 生成研究结论; 生成文章创新点 | 研究假设、结论、创新点,是基于科研人员的专业知识、洞察力和创造性思维,经过严格的研究过程产生的,不应使用AIGC工具生成。AIGC工具所生成的内容高度依赖语料与模型,其内容的专业性、逻辑性还需进行严格推敲 |
| 论文撰写 | 撰写整篇论文; 引用其他作者已标注为AIGC工具生成的内容 | 使用AIGC工具撰写整篇论文,存在违背学术伦理、内容不可靠等学术不端风险;国家自然科学基金委的《科研诚信规范手册》规定:不得将其他作者已标注为人工智能生成的内容作为原始文献引用,确需引用的应加以说明[19] |
表2 学术研究中不建议直接使用AIGC工具生成内容的场景Table 2 Scenarios where AIGC are not recommended for direct content generation in academic research |
| 阶段 | 不建议直接使用场景 | 注意事项 |
| 研究设计与研究开展 | 生成研究大纲; | AIGC模型基于大量的既有语料库训练,因而存在生成内容信息过时、偏颇等问题,并可能引发伦理和道德风险 |
| 描述详细的论文研究方法; | ||
| 阐述某些领域的专业知识和已有研究的概念; | ||
| 收集公众或专家的主观观点或带有情感倾向的信息; | ||
| 进行数据解释和背景说明; | ||
| 生成程序或代码 | AIGC模型的训练数据可能包含错误或者不准确的信息,其生成的内容可靠性无法保证。作者应对代码进行测试验证,避免出现错误、不完整或者低效的情况 |
表3 学术研究中可适当使用AIGC工具的场景Table 3 Scenarios where AIGC can be appropriately used in academic research |
| 阶段 | 可适当使用场景 | 注意事项 |
| 选题 | 启发选题思路 | 应确保数据和信息来源的可靠; 选题具有创新性,仍需发挥作者的思考和判断 |
| 研究设计与研究开展 | 辅助文献检索,确定关键词,扩展使用关键词,建立检索式; 检索、收集、分类和梳理相关文献; 辅助设计研究方法或验证假设的模型; 进行自动化数据清洗; 计算统计指标,描述统计结果; 生成问卷基础模板; 选刊推荐和投稿建议 | 作者需对文献来源、检索结果仔细核验,确保文献的真实性、相关性和质量; |
| 在研究方法的选取上避免因盲目依赖工具而忽视方法的适用性; | ||
| 作者需评估输入内容可能的安全风险并对数据清洗、计算统计结果进行审核把关; | ||
| 作者需结合实际情况进行可用性审核 | ||
| 论文撰写 | 总结主要内容,撰写摘要; | 学位论文或期刊有各自不同的写作和引用要求,作者需要仔细阅读并遵守的具体要求,对生成的内容进行仔细的审查和针对性的修改 |
| 提升内容的逻辑性,提供多样化的句式; | ||
| 实现多种语言的切换翻译; | ||
| 检查语法问题、拼写错误和润色论文; | ||
| 检查论文是否符合学术论文的格式; | ||
| 提供论文修改意见; | ||
| 生成符合学术论文引文格式要求的引用; | ||
| 对参考文献列表进行排序和顺序校对 |
张艳丽:确定研究思路,撰写论文初稿,修改论文与定稿;
管玉燕:相关政策资料收集整理,修改论文和校对格式;
宛 雪:调研、分析AIGC 工具,进行数据分析和解释,校对论文;
赵晶浩:调研、论证AIGC 工具生成原理和存在风险。
| [1] |
国家网信办,国家发展改革委,教育部,等.生成式人工智能服务管理暂行办法[EB/OL].[2023-12-20]. https://www.gov.cn/zhengce/zhengceku/202307/content_6891752.html.(Cyberspace Administration of China, National Development and Reform Commission, Ministry of Education, et al. Interim Measures for the Management of Generative Artificial Intelligence Services[EB/OL].[2023-12-20]. https://www.gov.cn/zhengce/zhengceku/202307/content_6891752.html.)
|
| [2] |
中国科学技术信息研究所,爱思唯尔,施普林格·自然,等.学术出版中AIGC使用边界指南[EB/OL].[2023-12-20]. https://www.istic.ac.cn/html/1/284/338/1701698014446298352.html.(Institute of Scientific and Technical Information of China, ELSEVIER, SPRINGER NATURE, et al. Guideline on the Boundaries of AIGC Usage in Academic Publishing[EB/OL].[2023-12-20].https://www.istic.ac.cn/html/1/284/338/1701698014446298352.html.)
|
| [3] |
和军,杨慧. ChatGPT类生成式人工智能监管的国际比较与借鉴[J].湖南科技大学学报(社会科学版),2023,26(6):119-128.(HE J,YANG H. International comparison and reference on ChatGPT-like Generative Artificial Intelligence Regulation[J].Journal of Hunan University of Science and Technology(Social Science Edition),2023,26(6):119-128.)
|
| [4] |
European Parliament. Artificial Intelligence Act[EB/OL].[2023-12-20].https://www.europarl.europa.eu/doceo/document/TA-9-2023-0236_EN.pdf.
|
| [5] |
闫宏秀,宋胜男.双重脆弱性与适度信任:从ChatGPT到Sora[J/OL].新疆师范大学学报(哲学社会科学版)[2024-04-21]. https://doi.org/10.14100/j.cnki.65-1039/g4.20240408.001.(YAN H X,SONG S N.Dual vulnerability and moderate trust: from ChatGPT to Sora [J/OL]. Journal of Xinjiang Normal University (Philosophy and Social Sciences Edition) [2024-04-21]. https://doi.org/10.14100/j.cnki.65-1039/g4.20240408.001.)
|
| [6] |
李艳. 三星ChatGPT泄密事件及启示 [J]. 保密工作, 2023 (9): 65-68. (LI Y. Samsung due to ChatGPT leakage incident and its inspiration [J]. Confidentiality, 2023, (09): 65-68.)
|
| [7] |
ZHAO R , LI X, CHIA Y K, et al. Can ChatGPT-like generative models guarantee factual accuracy? On the mistakes of new generation search engines[EB/OL].[2024-02-26]. https://arxiv.org/abs/2304.11076.
|
| [8] |
孙滔.发表3天被撤稿!中国作者用AI生成论文插图,每幅都荒谬[EB/OL].[2024-02-27].https://news.sciencenet.cn/htmlnews/2024/2/517714.shtm.(SUN T. Published for 3 days and was withdrawn! Chinese authors use AI to generate paper illustrations, and each one is absurd[EB/OL].(2024-02-20)[2024-02-27].https://news.sciencenet.cn/htmlnews/2024/2/517714.shtm.)
|
| [9] |
ELISABETH B. The rat with the big balls and the enormous penis – how Frontiers published a paper with botched AI-generated images[EB/OL].[2024-2-27].https://scienceintegritydigest.com/2024/02/15/the-rat-with-the-big-balls-and-enormous-penis-how-frontiers-published-a-paper-with-botched-ai-generated-images/.
|
| [10] |
种晓明. AIGC赋能出版业背景下版权监管新路径研究[J]. 科技与出版, 2023(8): 96-104.(ZHONG X M. A study on the new path of copyright supervision under the background of AIGC empowering the publishing industry[J]. Science-technology & publication, 2023 (8): 96-104.)
|
| [11] |
Nature. Why Nature will not allow the use of generative AI in images and video[EB/OL].[2023-12-20].https://www.nature.com/articles/d41586-023-01546-4.
|
| [12] |
THORP H H. ChatGPT is fun, but not an author[J].Science.2023,379(6630):313.
|
| [13] |
The lancet. Information for authors[EB/OL].[2023-12-20].https://www.thelancet.com/pb/assets/raw/Lancet/authors/tl-info-for-authors.pdf.
|
| [14] |
Elservier. The use of generative AI and AI-assisted technologies in writing for Elsevier[EB/OL].[2023-12-25].https://www.elsevier.com/about/policies-and-standards/the-use-of-generative-ai-and-ai-assisted-technologies-in-writing-for-elsevier.
|
| [15] |
Taylor & Francis Group. Taylor & Francis editorial policies on authorship[EB/OL].[2023-12-25].https://authorservices.taylorandfrancis.com/editorial-policies/defining-authorship-research-paper/.
|
| [16] |
ame5. arXiv announces new policy on ChatGPT and similar tools[EB/OL].[2023-12-25].https://blog.arxiv.org/2023/01/31/arxiv-announces-new-policy-on-chatgpt-and-similar-tools/.
|
| [17] |
王威.《布莱奇利宣言》:人工智能国际合作监管的新起点[J].服务外包,2023(12):46-52.(WANG W. Bletchley Declaration: a new starting point for international cooperation and regulation of artificial intelligence[J].China outsourcing,2023(12):46-52.)
|
| [18] |
科技部监督司.负责任研究行为规范指引(2023)[EB/OL].[2023-12-22].https://www.most.gov.cn/kjbgz/202312/t20231221_189240.html.(Supervision Department of the Ministry of Science and Technology. Guidelines for responsible research code of conduct (2023)[EB/OL].[2023-12-22].https://www.most.gov.cn/kjbgz/202312/t20231221_189240.html.)
|
| [19] |
国家自然科学基金委员会.科研诚信规范手册[EB/OL].[2023-12-22].https://www.nsfc.gov.cn/publish/portal0/tab442/info91294.html.(National Natural Science Fundation of China. Handbook of scientific research integrity standards[EB/OL].[2023-12-22].https://www.nsfc.gov.cn/publish/portal0/tab442/info91294.html.)
|
| [20] |
胡彬.人工智能可否被用来写医学论文[N].健康报,2024-01-17 (5). (Hu B.Can AI be used to write medical papers[N].Health, 2024-01-17 (5))
|
| [21] |
Silvia Vaccino-Salvadore.Exploring the Ethical Dimensions of using ChatGPT in language learning and beyond[J].Languages,2023,8(3):191.
|
| [22] |
MCADOO T. How to cite ChatGPT[EB/OL].[2024-01-29].https://apastyle.apa.org/blog/how-to-cite-chatgpt.
|
| [23] |
MLA style center. How do I cite generative AI in MLA style?[EB/OL].[2023-12-25]. https://style.mla.org/citing-generative-ai/.
|
| [24] |
全国信息与文献标准化技术委员会.信息与文献 参考文献著录规则: GB/T 7714-2015 [S].北京:中国标准出版社,2015.(Information and Documentation. Information and documentation—rules for bibliographic references and citations to information resources: GB/T 7714-2015[S].Beijing: Standard Press of China,2015)
|
/
| 〈 |
|
〉 |