Nonsense phrase "vegetative electron microscopy" spread through scientific papers via AI training data
A nonsensical scientific term "vegetative electron microscopy" originated from errors in digitizing 1950s papers and became embedded in AI training data, leading to its reproduction in at least 22 published papers. The phrase spread through plagiarized machine-generated text, particularly in papers from Iranian authors, and required retraction and correction actions by publishers.
What happened
The phrase 'vegetative electron microscopy' appeared in at least 22 scientific papers published across multiple journals including Springer Nature and Elsevier. The phrase originated from digitization errors in 1950s papers where 'vegetative' from one column was combined with 'electron' from another column, creating a nonsensical term.
The error was later incorporated into AI training data and subsequently reproduced in machine-generated scientific papers, particularly from Iranian authors. The phrase spread through plagiarism of AI-generated text containing the contaminated material.
Reported impact
The nonsense phrase became embedded in the scientific literature as what researchers call a 'digital fossil'βan error preserved and reinforced in AI systems that may be nearly impossible to remove from knowledge repositories. At least one paper containing the term was retracted by Springer Nature, while another from Elsevier received a correction.
Index analysis
The phrase served as a 'fingerprint' identifying paper mill involvement, according to scientific fraud investigators. A Russian chemist using the pseudonym 'Paralabrax clathratus' and software engineer Alexander Magazinov identified the term and speculated it originated from AI models trained on digitized historical papers.
Governance controls
Springer Nature retracted one paper containing the phrase. Elsevier issued a correction for another article whose senior author is an editor at the company, stating the publisher was 'content' with the wording.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.