由Taylor & Francis出版的同行评审期刊《新学术图书馆评论》发表的一篇关于Z-Library的论文被发现存在严重学术不诚实问题1。该论文在44个引用中至少有10个无法定位或根本不存在，包括一篇不存在的TorrentFreak文章《Publishers continue crackdown on shadow libraries》和一篇虚假的Harvard Journal of Law & Technology论文1。
论文致谢部分明确透露使用了Consensus AI进行文献筛选和信息提取，同时借助ChatGPT改进语言表述1。AI检测工具Pangram对该论文分析后报告，摘要和结论部分"100%可能是AI生成的"1，暗示这些虚假引用可能源于AI工具的幻觉。Taylor & Francis已将此事转交编辑和出版伦理诚信团队进行调查1。论文主作者及被引用的学者均未对后续提问做出回应1。
A peer-reviewed paper on Z-Library published in the New Journal of Library Reviews by Taylor & Francis has been found to contain at least ten non-existent references among its 44 total citations 1. These fabricated sources appear superficially credible but point to articles that were never actually published, including a purported TorrentFreak piece titled "Publishers continue crackdown on shadow libraries" and a non-existent Harvard Journal of Law & Technology article 1.
The paper's authors acknowledged in their acknowledgments section that they employed Consensus AI to sift through large volumes of literature and extract relevant information, and used ChatGPT to improve the language of their work 1. An AI detection tool called Pangram analyzed the paper and determined that the abstract and conclusion were "100% of the text is likely AI-generated," suggesting that artificial intelligence may be responsible for generating the false citations 1.
Following the discovery, Taylor & Francis has referred the matter to its editorial and publication ethics integrity team for investigation 1. The lead author has not responded to subsequent inquiries, nor have the cited scholars Gervais, Brembs, and Tenopir 1.
评论
还没有评论，欢迎留下第一条。