A recent analysis of 15 million biomedical abstracts shows that AI-generated content has become pervasive in academic publishing, particularly post-ChatGPT launch. Researchers utilized a method akin to tracking public health metrics to analyze language patterns in scientific writing. The study identifies a shift from concrete 'content words' to more 'stylistic and flowery' expressions, with some words showing increased frequencies. The research finds that at least 13.5% of papers published in 2024 involved AI assistance, showcasing a notable impact on academic writing.
The study reveals that since ChatGPT's release, academic papers show a shift from concrete 'content words' to 'stylistic and flowery' language, indicating AI's influence.
The analysis tracked 'excess words' in scientific writing before and after large language models arrived, showing a significant infiltration of AI-generated content.
Collection
[
|
...
]