A recent study has found that the rise of large language models (LLMs) like ChatGPT has significantly impacted scientific writing in biomedical research. Researchers analyzed over 15 million biomedical abstracts from the PubMed library and discovered a marked increase in “flowery” language, with verbs such as “delves,” “underscores,” and “showcasing” being used more frequently.
The study’s findings suggest that LLMs have led to abrupt changes in both the quality and quantity of research papers. The analysis revealed that at least 13.5% of abstracts published last year were processed with LLMs, equating to around 200,000 papers. Co-author Ágnes Horvát warned that the impact of ChatGPT could have broader implications for scientific writing, including making up references, providing inaccurate summaries, and making false claims that sound authoritative.
The researchers caution that the increased use of LLMs may lead to a homogenization of scientific writing, resulting in less diverse and novel papers. They call for a reassessment of current policies and regulations surrounding the use of LLMs in science.
Source: https://www.timeshighereducation.com/news/chatgpt-drives-rise-flowery-language-journal-abstracts