{"id":13341,"date":"2026-05-17T21:38:07","date_gmt":"2026-05-18T01:38:07","guid":{"rendered":"https:\/\/www.danielharper.org\/yauu\/?p=13341"},"modified":"2026-05-18T21:48:17","modified_gmt":"2026-05-19T01:48:17","slug":"noted-with-minimal-comment-5","status":"publish","type":"post","link":"https:\/\/www.danielharper.org\/yauu\/2026\/05\/noted-with-minimal-comment-5\/","title":{"rendered":"Noted with minimal comment"},"content":{"rendered":"\n

Excerpt from Maxim Topaz, Nir Roguinb, Pallavi Guptab, Zhihong Zhanga, and Laura-Maria Peltonenf,\u201cFabricated citations: an audit across 2\u00b75 million biomedical papers,\u201d Correspondence, The Lancet,<\/em> vol. 407, issue 10541, P1779-1781, 9 May 2026<\/a>:<\/p>\n\n\n\n

\u201cScientific literature depends on the integrity of its references. Each reference implicitly asserts that a verifiable source exists and supports the claims being made. When references point to non-existent studies, readers, reviewers, and policy makers are unable to evaluate the evidence.<\/p>\n\n\n\n

\u201cFabricated references (references whose claimed titles correspond to no existing publication) can arise from paper mill activity, intentional misconduct, or uncritical use of artificial intelligence (AI) writing tools.\u00a0Large language models (LLMs) generate plausible sounding but fictitious references, a well documented failure mode; previous studies estimate that 30\u201369% of LLM-generated references in biomedical contexts are fabricated.\u00a0These references are often correctly formatted, attributed to real researchers, and bear plausible publication dates, making them difficult to detect by conventional peer review.\u00a0To our knowledge, no systematic audit of reference integrity across the biomedical literature has been conducted until now.<\/p>\n\n\n\n

\u201cWe present findings from a reference-integrity audit of 2\u00b75 million biomedical papers spanning 3 years, showing that fabricated references are embedded in the peer-reviewed literature at scale, and that the rate of fabrication is accelerating….<\/p>\n\n\n\n

\u201cIn 2023, approximately one in 2828 papers contained at least one fabricated reference. By 2025, this had risen to one in 458 and in the first 7 weeks of 2026, one in 277 papers had at least one fabricated reference. The fabrication rate increased more than 12 times, from approximately four per 10?000 papers in 2023, to 51\u00b73 per 10?000 papers in the fourth quarter of 2025, reaching 56\u00b79 per 10?000 papers in early 2026….\u201d<\/p>\n\n\n\n

In a note at the end, the authors state that they used generative AI: \u201cDuring the preparation of this work the authors used Claude (Anthropic) in order to assist with code development, grammar, and punctuation. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.\u201d<\/p>\n\n\n\n

So the message here is not<\/em> \u201cdon’t use generative AI.\u201d The message here is: \u201cIf you use generative AI, you need to know its limitations, and you need to take responsibility for things like fact-checking, checking references, etc.\u201d In short, when you use AI, you still have to take full responsibility for whatever AI produces.<\/p>\n","protected":false},"excerpt":{"rendered":"

Excerpt from Maxim Topaz, Nir Roguinb, Pallavi Guptab, Zhihong Zhanga, and Laura-Maria Peltonenf,\u201cFabricated citations: an audit across 2\u00b75 million biomedical papers,\u201d Correspondence, The Lancet, vol. 407, issue 10541, P1779-1781, 9 May 2026: \u201cScientific literature depends on the integrity of its references. Each reference implicitly asserts that a verifiable source exists and supports the claims being … Continue reading “Noted with minimal comment”<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[84],"tags":[940],"class_list":["post-13341","post","type-post","status-publish","format-standard","hentry","category-new-media","tag-applied-statistics"],"_links":{"self":[{"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/posts\/13341","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/comments?post=13341"}],"version-history":[{"count":1,"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/posts\/13341\/revisions"}],"predecessor-version":[{"id":13342,"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/posts\/13341\/revisions\/13342"}],"wp:attachment":[{"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/media?parent=13341"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/categories?post=13341"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.danielharper.org\/yauu\/wp-json\/wp\/v2\/tags?post=13341"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}