AI models fed AI-generated data quickly spew nonsense
Open accessAbstract
Researchers gave successive versions of a large language model information produced by previous generations of the AI — and observed rapid collapse. Researchers gave successive versions of a large language model information produced by previous generations of the AI — and observed rapid collapse.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Concerns persist regarding the reliability of generative AI, particularly in ensuring factual correctness in medical contexts.
“However, despite these advantages, concerns persist regarding its reliability, particularly in ensuring factual correctness in medical contexts.”
From Integrating AI in medical education: a comprehensive study of medical students’ attitudes, concerns, and behavioral intentions (2025) · cites this paper as [25]
Cites
AI models collapse when trained on recursively generated data
Shumailov, I., Shumaylov, Z., Zhao, Y., Papernot, N., Anderson, R., & Gal, Y. (2024) · Nature · doi:10.1038/s41586-024-07566-y · Cited by 793 · see citing works
Beyond Model Collapse: Scaling Up with Synthesized Data Requires VerificationPreprint
Feng, Y., Dohmatob, E., Yang, P., Charton, F., & Kempe, J. (2024) · International Conference on Learning Representations · doi:10.48550/arxiv.2406.07515 · Cited by 54 · see citing works
Gerstgrasser, M., Schaeffer, R., Dey, A., Rafailov, R., Sleight, H., Hughes, J., Korbak, T., Agrawal, R., Pai, D., Gromov, A., et al. (2024) · arXiv · doi:10.48550/arxiv.2404.01413 · Cited by 15 · see citing works
Nepotistically Trained Generative-AI Models CollapsePreprint
Boháček, M., & Farid, H. (2023) · arXiv (Cornell University) · doi:10.48550/arxiv.2311.12202 · Cited by 5 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Duan, S., Liu, C., Rong, T., Zhao, Y., & Liu, B. (2025) · BMC Medical Education · doi:10.1186/s12909-025-07177-9 · Cited by 85 · see citing works
“AI-generated content could exhibit mistakes, nonsensical elements, insufficient sourcing, or references to non-existent sources [25, 26].”
Generative AI and Empirical Research Methods in Operations Management
Shalpegin, T., Browning, T. R., Kumar, A., Shang, G., Thatcher, J. B., Fransoo, J. C., Holweg, M., & Lawson, B. (2025) · Journal of Operations Management · doi:10.1002/joom.1371 · Cited by 12 · see citing works
Petroff, D. J., Nasir, A. A., Moin, K., Loveless, B., Moshirfar, O. A., Hoopes, P. C., & Moshirfar, M. (2024) · Cureus · doi:10.7759/cureus.67747 · Cited by 6 · see citing works
Leveraging AI for Advancements in Qualitative Research Methodology
Haouam, I. (2025) · Journal on artificial intelligence · doi:10.32604/jai.2025.064145 · Cited by 10 · see citing works
Petrophysical prediction of oil reservoirs using explainable artificial intelligence
Li, S., Wei, Z., Yang, Y., Zhang, J., & Cheng, Y. (2025) · Geophysics · doi:10.1190/geo2024-0271.1 · Cited by 4 · see citing works
Human-inspired Perspectives: A Survey on AI Long-term Memory
He, Z., Lin, W., Zheng, H., Zhang, F., Jones, M. W., Aitchison, L., Xu, X., Liu, M., Kristensson, P. O., & Shen, J. (2024) · arXiv · doi:10.48550/arxiv.2411.00489 · Cited by 3 · see citing works
“Recursive use of model-generated texts as training data can precipitate model collapse [170, 171].”
GeoAI collapse? Ethical implications of synthetic geospatial data use
Zhao, B., & Lin, Y. (2026) · International Journal of Geographical Information Systems · doi:10.1080/13658816.2025.2609806 · Cited by 3 · see citing works
An umbrella review encompassing 42 systematic reviews on medical applications of ChatGPT
Ramadan, S., Calvieri, C., Gracia-Ramos, A. E., FERRARI, G. R., Pepe, M., & BIONDI-ZOCCAI, G. (2026) · Minerva Medica · doi:10.23736/s0026-4806.25.09664-8 · Cited by 2 · see citing works
Dimensionality-modulated generative AI for safe biomedical dataset augmentation
Lötsch, J., Himmelspach, A., & Kringel, D. (2025) · iScience · doi:10.1016/j.isci.2025.114321 · Cited by 1 · see citing works
Zhao, L., Han, Q., Li, P., Dai, H., & Ju, X. (2026) · BMC Medical Education · doi:10.1186/s12909-026-08947-9 · Cited by 1 · see citing works
From prompt engineering to prompt design: Research strategies for visual generative AI
Colombo, G., Niederer, S., & Gaetano, C. D. (2026) · Big Data & Society · doi:10.1177/20539517261451462 · Cited by 3 · see citing works
Rogers, E., Turbé, V., Gareta, D., Herbst, C., McKendry, R. A., Herbst, K., & Shahmanesh, M. (2025) · medRxiv · doi:10.1101/2025.02.25.25322677 · Cited by 1 · see citing works
Lötsch, J., & Kringel, D. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5366515
Navigating creativity and AI: Possibilities, precautions, and perspectives
Wilson, J., Worwood, M. J., & Kaufman, J. C. (2025) · Elsevier eBooks · doi:10.1016/b978-0-443-34073-4.00014-9
May AI be sustainable? An outlook on sustainability and technology
Kondratenko, A., Kondratenko, A. A., & Kamberov, K. H. (2025) · AI & Society · doi:10.1007/s00146-025-02804-x
Skills Training in the Age of AI
Blakely, C., & Rose, R. L. (2026) · Springer eBooks · doi:10.1007/978-981-95-1622-3_5
Will AI accelerate or undermine the way humans have always innovated?
Bentley, R. A. (2026) · doi:10.64628/aai.mjsrjrqw3
Chen, Y., Lei, M., Liu, Z., Tang, M., & She, J. (2026) · Computers in Human Behavior · doi:10.1016/j.chb.2026.108967
Mammino, L. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-13417-3.00007-x
- Authors
- Gibney, E.
- Corresponding author
- Gibney, Elizabeth
- Venue
- Nature
- Published
- 2024-07-24
- Volume
- 632
- Issue
- 8023
- Pages
- 18-19
- ISSN
- 0028-0836 (ISSN-L)
- 1476-4687
- OpenAlex
- W4400944866
- Cited by
- 18
- References
- 4
- Citations by year
- 2026: 8 · 2025: 9 · 2024: 1
- Type
- other
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- Locations
- Nature (publishedVersion, cc-by)
- PubMed (publishedVersion)
- RePEc: Research Papers in Economics (submittedVersion)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Keywords
- large language models, AI-generated data, self-training, catastrophic collapse, data quality
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
