So they established that language patterns measured by word frequency changed between 2022 and 2024. But did they also analyse frequencies across other 2-year time periods? How much difference is there for a typical word? It looks like they have a per-frequency significance threshold but then analysed all words at once, meaning that random noise would turn up a bunch of "significant" results. Maybe this is addressed in the original paper which is not linked.

direct link to the paper, rather than this Gazeon clickbait: https://www.science.org/doi/10.1126/sciadv.adt3813

We study vocabulary changes in more than 15 million biomedical abstracts from 2010 to 2024 indexed by PubMed and show how the appearance of LLMs led to an abrupt increase in the frequency of certain style words. This excess word analysis suggests that at least 13.5% of 2024 abstracts were processed with LLMs. This lower bound differed across disciplines, countries, and journals, reaching 40% for some subcorpora.