A novel approach to detecting large language model (LLM) usage in scientific writing has been developed by a team of researchers, who analyzed 14 million paper abstracts published on PubMed between 2010 and 2024. By tracking the relative frequency of certain words, they found that at least 10% of 2024 abstracts were processed with LLMs, with a surge in usage of “style words” such as verbs, adjectives, and adverbs. The researchers identified hundreds of “marker words” that became significantly more common in the post-LLM era, including words like “delves,” “showcasing,” and “underscores.” This method provides a unique insight into the impact of LLMs on scientific writing and raises concerns about the potential misuse of AI-generated text.
Researchers Crack Code to Detect AI-Generated Writing in Scientific Papers
The researchers found hundreds of words with sudden, pronounced increases in scientific usage that had no common link to world events.
1–2 minutes
TOP STORIES
Anthropic continues to engage with the Trump administration despite Pentagon tensions …
Lawmakers express concerns about AI’s rapid evolution and its risks …
OpenAI is losing key leaders as it shifts focus to enterprise AI and its superapp …
Maine’s new bill pauses large data center construction to assess environmental impacts …
Authorities arrested Daniel Moreno-Gama for attacking OpenAI CEO Sam Altman over his fears about AI …
Anthropic’s Mythos model raises national security concerns while sparking a lawsuit against the DOD …
