(10 Sept 2025) Publishers increasingly require and academics are increasingly willing to acknowledge and credit the role played by AI in their research papers. Discussing a recent study, of these acknowledgements Kayvan Kousha finds LLMs to be used for a fairly narrow set of academic tasks.
There is growing evidence that many researchers are using LLMs. Surveys suggest that a significant number already rely on tools like ChatGPT, primarily for writing, reviewing, brainstorming, or translation. For instance, 31% of researchers (n=2,284) used LLMs for research tasks; 93% said it helped with writing or reviewing. In another survey, 55% of researchers (n=1,659) said LLMs saved time and money; mainly used for writing/editing, especially by non-native English speakers. Other studies indicate that AI-related terms have appeared more frequently in academic publications and reviews since the release of ChatGPT. In another similar recent study on the influence of LLMs on academic publishing myself and Michael Thelwall found that some LLM-associated terms showed dramatic growth of 400–1600% between 2022 and 2024, suggesting that certain terms will naturally become more common in academic writing.
Analyzing acknowledgments in academic papers offers a more direct way to assess how researchers are specifically using LLMs in their publications. This study analyzed 1,759 papers indexed in Scopus and Web of Science up to August 2024 that explicitly mentioned ChatGPT in their acknowledgments.
Read more here.
