Nazarovets, S. та Teixeira da Silva, J. A. (2026) Can Generative AI Assist Post-publication Peer Review? A Pilot Study on LLM-assisted Triage Journal of Academic Ethics, 24 (86). ISSN 1570-1727
Повний текст недоступний з цього архіву.Анотація
Generative artificial intelligence (GenAI) conversational agents such as ChatGPT and DeepSeek are increasingly discussed as tools that may influence different stages of scholarly communication. While much of the recent debate has focused on their use in writing and editing, considerably less attention has been paid to whether such systems might assist post-publication peer review (PPPR), which plays an important role in identifying errors, inconsistencies, and potential integrity concerns in the published literature. This pilot study explores whether GenAI tools can assist PPPR by flagging potentially problematic elements in scientific papers, acting as triage tools rather than autonomous detection systems. Three previously documented integrity-related cases were used to examine whether the models would raise signals of concern when presented with: (1) text containing “tortured phrases”, (2) text from a paper later retracted due to integrity-related concerns, and (3) a publication later corrected for a figure-related issue. In a second exercise, the study assessed whether the models could assist in screening for potentially problematic citations, including references to retracted publications, in the reference lists of review papers. The results indicate that the tested systems did not consistently raise concerns corresponding to known integrity-related problems and showed limited reliability in identifying potentially retracted references. While DeepSeek occasionally noted linguistic or structural irregularities, neither system consistently raised concerns that corresponded to the documented issues in the tested papers. These findings suggest that, in their current form, these GenAI tools are better understood as preliminary screening assistants that help draw attention to elements requiring further human evaluation. Conversely, they should not be interpreted as systems capable of reliably verifying the integrity of the scientific record or independently establishing the presence of integrity-related problems without external databases and expert assessment.
| Тип елементу : | Стаття |
|---|---|
| Ключові слова: | Artificial intelligence; Citations; Indexing; Integrity; Large language models; Metrics; Peer review; Replication; Retractions; Transparency |
| Типологія: | Статті у базах даних > Scopus > У виданнях Q1 Scopus Статті у базах даних > Web of Science |
| Підрозділи: | Бібліотека |
| Користувач, що депонує: | Сергій Андрійович Назаровець |
| Дата внесення: | 02 Вер 2026 15:27 |
| Останні зміни: | 02 Вер 2026 15:27 |
| URI: | https://elibrary.kubg.edu.ua/id/eprint/59455 |
Actions (login required)
 |
Перегляд елементу |