# Post Analysis: @thesupermannx on "model collapse" (2026-09-26) Back to [[Users/@thesupermannx/thesupermannx-x-profile|@thesupermannx profile]] · Paper note: [[Users/@thesupermannx/Papers/AI Models Collapse on Recursively Generated Data (Nature 2024)|Nature 2024]] · Related pattern: [[Topics/Ribosome Pausing and Cotranslational Folding/Related Trends#1. "AI cracked open X" posts that re-label old science|Ribosome topic: the "AI cracked open X" pattern]] **Verdict:** Mostly accurate: the post closely paraphrases the 2024 Nature abstract, but presents a two-year-old paper as breaking news ('ChatGPT has now a big problem'). ## The post (verbatim) Posted **Sep 26, 2026, 6:13 AM CT**. Link: [PubMed 39048682](https://pubmed.ncbi.nlm.nih.gov/39048682/). > ChatGPT has now a big problem. > > Researchers at Oxford and Cambridge exposed a massive threat to large language models.” > > They call it “model collapse." > > Internet ecosystem is rapidly changing, and generative AI will soon contribute much of the text found online. This forces us to consider what happens to future iterations like gpt-n when they are trained on data scraped from the web that was already generated by an llm. > > According to the research, indiscriminately using model-generated content in training causes "irreversible defects" in the resulting ai. the model loses the "tails of the original content distribution." in other words, it begins to forget the creative, fringe, and unique nuances of actual human writing, collapsing into a repetitive echo chamber. > > This isn't just a chatgpt issue.. the researchers built theoretical intuition showing this collapse is ubiquitous across learned generative models, occurring in large language models as well as in variational autoencoders and gaussian mixture models. > > Tech companies rely on scraping the internet for large-scale data to build smarter models. However, the paper warns that if we want to sustain the benefits of training on web data, model collapse must be taken seriously. > > Ultimate takeaway? data collected from genuine human interactions is going to become increasingly valuable in a web filled with ai content. ## Engagement (Oct 24, 2026) | Impressions | Likes | Reposts | Quotes | Replies | Bookmarks | |---:|---:|---:|---:|---:|---:| | 1,316,316 | 3,581 | 1,092 | 360 | 412 | 2,555 | ## Claim-by-claim check **Ratings:** ✅ accurate · 🟡 overstated or simplified · 🔀 misattributed · ❓ unsupported by the paper · ❌ contradicted by the paper | # | Claim | Rating | What the sources show | |---|---|---|---| | 1 | "ChatGPT has now a big problem" (implied news) | 🟡 | The paper was published Jul 24, 2024, 26 months before the post | | 2 | "Researchers at Oxford and Cambridge" | 🟡 ⚠️ | Six authors. Affiliations were not confirmed this run, and the list is reported to span more institutions than these two | | 3 | Indiscriminate use of model-generated content causes "irreversible defects"; tails disappear | ✅ | Close paraphrase of the abstract | | 4 | Ubiquitous across LLMs, VAEs and GMMs | ✅ | Abstract | | 5 | Model collapse "must be taken seriously"; human-interaction data will grow in value | ✅ | Abstract | ## What the post leaves out - **The paper's age:** it is from 2024. - **The follow-up debate:** later work disputed how inevitable collapse is when real data is kept alongside synthetic data. One example is Gerstgrasser et al., "Is Model Collapse Inevitable?" ([arXiv 2404.01413](https://arxiv.org/abs/2404.01413)) ⚠️ (not re-read this run). ## Bottom line - **The text is reliable:** it is largely lifted from the abstract. - **The framing is the problem:** an old paper is presented as a fresh crisis. - **Other old papers presented as new:** - [Natural Selection Favors AIs over Humans](https://x.com/thesupermannx/status/2104241247523410014) (arXiv 2303.16200, 2023) - [ElectionSim](https://x.com/thesupermannx/status/2092823839453626714) (arXiv 2410.20746, 2024)