A new study published on arXiv has revealed that large language models are contributing to the problem of "phantom references" or hallucinated citations in scientific papers. These fabricated citations, which do not resolve to real scholarly works, are appearing in peer-reviewed proceedings from top-tier conferences like ICLR, ICML, NeurIPS, and USENIX Security. The research indicates that while overall rates are low, the sheer volume of papers means a significant number contain these errors, with some papers exhibiting multiple hallucinations. The study also noted an increase in such errors following the release of ChatGPT, even appearing in award-winning papers, suggesting that current peer review processes are insufficient to catch these AI-generated inaccuracies. AI
IMPACT Highlights a critical flaw in AI-generated scientific text, potentially undermining research integrity and necessitating new verification methods.
RANK_REASON Research paper detailing a novel finding about AI-generated content in academic publications. [lever_c_demoted from research: ic=1 ai=1.0]
- arXiv
- ChatGPT
- Conference on Neural Information Processing Systems
- International Conference on Learning Representations
- International Conference on Machine Learning
- RefChecker
- USENIX Security Symposium
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
