A recent study analyzing over 200,000 astronomy papers found that more than half of them show traces of language model assistance. The research, which modeled the presence of distinctive language model vocabulary, estimated that 54% of papers from 2025 exhibited such assistance, though only 0.81% of those papers disclosed their use of AI tools. The study also noted that the linguistic markers of AI assistance are becoming less apparent over time as authors adapt their writing. AI
IMPACT Indicates a significant shift in scientific writing practices, raising questions about authorship and originality in research.
RANK_REASON Research paper published on arXiv detailing findings about AI assistance in scientific writing.
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →