A study from the University of Tübingen highlights that AI has significantly affected scientific vocabulary, with increased usage of words like 'delves,' 'showcasing,' 'underscores,' and 'potential,' which are favored by AI models.

Analysis of over 15 million biomedical abstracts from 2010 to 2024 reveals that at least 13.5% of papers in 2024 involved AI assistance, with some fields and countries showing up to 40% adoption.

The stylistic changes include a reduction in content-heavy nouns and an increase in verbs and adjectives, reflecting AI-driven modifications in scientific language.

In computational fields, about 20% of abstracts involved AI, likely due to greater familiarity among computer scientists, while in non-English speaking countries, AI may assist with English editing.

Post-ChatGPT release, there has been a noticeable shift from noun-heavy to verb- and adjective-heavy language, indicating AI's role in shaping scientific vocabulary and style.

Further, the research indicates that around 13% of biomedical abstracts published in 2024 used large language models, reflecting a widespread integration of AI in scientific writing.

The study also uncovered significant variability in AI usage across disciplines, countries, and publication venues, demonstrating diverse adoption patterns worldwide.

Researchers have observed a notable shift in scientific writing style post-2024, with increased use of flowery and stylistic language such as 'showcasing,' 'pivotal,' and 'grappling,' indicating AI's influence on academic language.

Some researchers openly acknowledge AI assistance, such as a radiology paper explicitly stating it was written by a chatbot, while others attempt to conceal AI use by altering their language.

Using an 'excess word' framework, researchers estimate that AI's impact on scientific language exceeds even major events like COVID-19, marking an unprecedented shift.

Language pattern analysis identified 454 words frequently used by large language models like ChatGPT, serving as markers for AI involvement in scientific texts.

The primary influence of AI appears to be on the tone and style of writing, with a shift toward more affective verbs and adjectives rather than content-related nouns.