Cinta Gallent, co-author of a study that detects an accelerated change in the vocabulary of scientific articles since 2022, coinciding with the advent of generative AI.
New study published by Learned Publishing in Wiley Online Library, written by Rubén Comas Forgas, Cinta Gallent Torres, Carmen Touza Garma and Alex Glynn, finds that the advent of generative artificial intelligence tools is changing the way academic writing is done. The research analyses millions of abstracts of scientific articles indexed in Scopus and identifies an accelerated increase, from 2022, of certain words and phrases associated with the use of AI, coinciding with the popularisation of tools such as ChatGPT.
Using a combination of bibliometric and content analysis techniques, the researchers identified 17 terms that experienced unusual growth between 2022 and 2024. The study examined their diffusion across different scientific disciplines, countries and journals, considering them as probabilistic indicators of a linguistic shift linked to the AI era, but not as direct proof that a paper has been written with these tools.
The results show a clear acceleration in the incorporation of this vocabulary after 2022, consistent with AI-assisted writing environments. However, the authors note that this phenomenon is likely due to several factors, including both the direct use of AI and the imitation of styles, editorial normalisation, or a general evolution of academic language.
The study also highlights a significant difference between the number of articles that explicitly declare having used AI tools and the prevalence of these linguistic patterns. The researchers interpret this discrepancy as a possible asymmetry in the way the use of these technologies is reported, rather than as evidence of undeclared use.
The findings suggest that artificial intelligence is significantly transforming written expression in scientific communication and raise the need to establish clearer criteria on the declaration of AI use, reinforce editorial guidelines, and promote a deeper debate on authorship, scientific integrity, and linguistic diversity in academic production.
