LLMMedium
Reported Emergence of 'Vegetative Electron Microscopy' in Scientific Papers Traced to Purported AI Training Data Contamination
- Occurred
- Apr 15, 2025
- Published
- Feb 10, 2025
- Source
- AI Incident Database
Summary
Researchers reportedly traced the appearance of the nonsensical phrase "vegetative electron microscopy" in scientific papers to contamination in AI training data. Testing indicated that large language models such as GPT-3, GPT-4, and Claude 3.5 may reproduce the term. The error allegedly originated from a digitization mistake that merged unrelated words during scanning, and a later translation error between Farsi and English.
Developers
- OpenAI
- Anthropic
Deployers
- OpenAI
- Anthropic
- Researchers
- Scientific authors
Harmed parties
- Researchers
- Scientific authors
- Scientific publishers
- Peer reviewers
- Scholars
- Readers of scientific publications
- Scientific record
- Academic integrity
Tags
- training data contamination
- knowledge integrity
- hallucination
- scientific publishing
- large language models
- epistemic harm
- digitization error
- translation error
- academic integrity
Classification
Category confidence: 90%
Severity confidence: 85%
Rationale: AI models reproducing nonsensical terms due to training data contamination represents a notable LLM failure.
Classifier: Qwen/Qwen3.5-35B-A3B-FP8