摘要
生成式人工智能凭借强大的语言理解与生成能力,在众多领域展现出了巨大潜力,并在科研工作中得到广泛应用,提升了科学研究的整体效率。但随着生成式人工智能工具的深入应用,其在科研领域的潜在风险也日益凸显。文章阐述了生成式人工智能在科研中的应用状况,探讨了其参与科研时在文献提取、原创性、内容撰写、图像生成、审稿质量、价值判断等方面可能引发的风险,这些风险影响了科研成果的质量与产出,加剧学术不端行为,对学术诚信构成严重威胁,并可能引发信息安全问题。文章指出了科技期刊必须要高度警觉生成式人工智能技术带来的潜在风险,及时采取有效应对策略,以提高对人工智能生成内容(AIGC)的识别和处理能力,确保生成式人工智能工具得到负责任的使用,维护科研诚信,保障学术出版的严谨性与科学性。同时,文章强调了目前AI不具备价值判断等能力,并提出相应建议明确人类在科研活动中的最终裁量权与责任主体地位。
Generative artificial intelligence,with its powerful language understanding and generation capabilities,has demonstrated tremendous potential in numerous fields and has been widely applied in scientific research,enhancing the overall efficiency of scientific endeavors.However,as generative AI tools becomes more deeply integrated,its potential risks in the field of scientific research are also increasingly prominent.This paper elaborated on the application status of generative AI in scientific research and discussed the risks that may arise when it is involved in scientific research,including issues related to literature extraction,originality,content writing,image generation,peer review quality,and value judgment.These deficiencies impact the quality and output of scientific research results,exacerbate academic misconduct,pose a serious threat to academic integrity,and may even lead to information security issues.This paper pointed out that scientific journals must be highly vigilant about the potential risks posed by generative artifici al intelligence(GAI)technology and should promptly adopt effe ctive countermeasures.Doing so will enhance their capability to identify and handle AI-generated content(AIGC),ensure the responsible use of generative AI tools,maintain scientific integrity,and safeguard the rigor and scientific quality of academic publishing.It also emphasized that current AI lacks the capacity for value judgment and social context awareness,proposing corresponding recommendations to clarify that ultimate discretion and primary responsibility in research activities ultimately rest with humans.
作者
王丽娜
Wang Lina(Editorial Department of Science and Technology Review,Science and Technology Review Publishing House,Beijing 100081,China)
出处
《今日科苑》
2025年第4期47-56,共10页
Modern Science
关键词
生成式人工智能
人工智能生成内容
大语言模型
科技期刊
潜在风险与应对策略
Generative Artificial Intelligence(GAI)
AI-Generated Content(AIGC)
Large Language Models(LLMs)
scientific journals
potential risks and countermeasures
