Researchers have found that when AI language models hallucinate citations, the fabricated author names are not random. Instead, these models generate correlated ensembles of co-authors whose co-occurrence rates are higher than chance and consistent across generations. These fabricated records, which include non-existent journals and backdated publication dates, are assigned real DOIs, allowing them to be ingested by scholarly aggregators. AI
IMPACT Reveals systematic patterns in AI-generated fake citations, potentially impacting scholarly integrity and AI detection methods.
RANK_REASON The cluster describes findings from a research paper about AI hallucination patterns. [lever_c_demoted from research: ic=1 ai=1.0]
Read on Mastodon — fosstodon.org →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
