Department of Orthopaedic Surgery, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts.
Department of Orthopaedic Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts.
J Bone Joint Surg Am. 2023 Sep 6;105(17):1388-1392. doi: 10.2106/JBJS.23.00473. Epub 2023 Jul 12.
➢ Natural language processing with large language models is a subdivision of artificial intelligence (AI) that extracts meaning from text with use of linguistic rules, statistics, and machine learning to generate appropriate text responses. Its utilization in medicine and in the field of orthopaedic surgery is rapidly growing.➢ Large language models can be utilized in generating scientific manuscript texts of a publishable quality; however, they suffer from AI hallucinations, in which untruths or half-truths are stated with misleading confidence. Their use raises considerable concerns regarding the potential for research misconduct and for hallucinations to insert misinformation into the clinical literature.➢ Current editorial processes are insufficient for identifying the involvement of large language models in manuscripts. Academic publishing must adapt to encourage safe use of these tools by establishing clear guidelines for their use, which should be adopted across the orthopaedic literature, and by implementing additional steps in the editorial screening process to identify the use of these tools in submitted manuscripts.
➢ 基于大型语言模型的自然语言处理是人工智能(AI)的一个分支,它使用语言规则、统计学和机器学习从文本中提取意义,以生成合适的文本回复。它在医学和骨科领域的应用正在迅速发展。➢ 大型语言模型可用于生成具有出版质量的科学手稿文本;然而,它们存在 AI 幻觉,其中会以误导性的信心陈述虚假或半真半假的内容。它们的使用引发了人们对潜在研究不当行为以及幻觉将错误信息插入临床文献的严重关注。➢ 当前的编辑流程不足以识别手稿中是否使用了大型语言模型。学术出版必须适应通过为其使用制定明确的指南来鼓励安全使用这些工具,这些指南应该在整个骨科文献中采用,并通过在编辑筛选过程中实施额外的步骤来识别提交的手稿中使用这些工具的情况。