М. Ю. Нещерет The article is devoted to the problem of false bibliographic records and links fabricated by a chatbot with generative artificial intelligence. Text generators have become popular due to the ability to save time and automate the content creation process. Using arrays of heterogeneous data, artificial intelligence is able to combine and interpret various sources of information and create new texts based on them. For bibliographic records, which are a special type of text, any paraphrasing becomes disastrous, because elements of a bibliographic record are unique identifiers of a specific information resource. Spreading of fabricated bibliographic records and links to non-existent publications indicates that artificial intelligence is not able to accurately recognize which parts of the text should be considered as bibliographic data, which significantly limits the possibilities of using chatbots in scientific research. Over time, as technology improves, artificial intelligence must learn to identify bibliographic records and process them differently from ordinary text – copy them exactly, not paraphrase them. However, until these problems are resolved, researchers should exercise caution: information obtained via the use of ChatGPT should be verified. The author of the article focuses on the importance of creating preventive measures to prevent the dissemination of unreliable bibliographic records and references, since this can lead to serious failures in the functioning of the bibliographic information system.
Научная статья 
генеративный чат-бот