#ModelCollapse
LightNews LightNews
bobehayes.bsky.social
#ChatGPT Has Already Polluted the Internet So Badly That It's Hobbling Future AI Development

futurism.com/chatgpt...

#AI #ModelCollapse
ChatGPT Has Already Polluted the Internet So Badly That It's Hobbling Future AI Development
There may be no undoing the vast amounts of pollution wreaked by ChatGPT. And that's just tough luck for any AI models that come after it.
futurism.com
June 18, 2025 at 10:17 AM
greybe.bsky.social
In summary, Model Collapse presents significant challenges for AI development, reliability, and sustainability. As we move forward, addressing these issues will be crucial for maintaining trust and advancing technology responsibly. #ModelCollapse #AI #FutureOfAI
December 6, 2024 at 12:30 AM
mryeswecan.bsky.social
Comparison: AI #modelcollapse and ideological collapse of Silicon Valley’s elites. Just as AI models degrade when trained on low-quality, self-referential data, figures like Musk, Thiel, and Andreessen are spiralling into a bubble of their own making.

www.aaronrosspowell.com/p/silicon-va...
Silicon Valley’s Very Online Ideologues are in Model Collapse
Like an AI trained on its own output, they’re growing increasingly divorced from reality, and are reinforcing their own worst habits of thought.
www.aaronrosspowell.com
October 30, 2024 at 9:26 AM
danigambit.bsky.social
Anyone here on #bluesky studying #LLMs? Here's our preprint about #autophagy and linguistic analysis of #ModelCollapse we are working on... #ai #phd #research
arxiv.org/abs/2410.12341
A linguistic analysis of undesirable outcomes in the era of generative AI
Recent research has focused on the medium and long-term impacts of generative AI, posing scientific and societal challenges mainly due to the detection and reliability of machine-generated information...
arxiv.org
November 15, 2024 at 11:36 AM
technoholic.bsky.social
🤖 "All human data for AI training is exhausted," says Musk. Is synthetic data the future or a recipe for disaster? Experts weigh in!

#AIConcerns #ModelCollapse #ElonMusk
January 10, 2025 at 2:23 PM
drimplausible.com
Or as we like to call it: "Soylent Culture" #ai #chatgpt #soylentculture #modelcollapse

futurism.com/ai-models-fa...
AI Models Show Signs of Falling Apart as They Ingest More AI-Generated Data
As CEOs trip over themselves to invest in AI, the models are falling apart at the seams and going mad from cannibalism.
futurism.com
June 11, 2025 at 5:28 PM
danigambit.bsky.social
Our new #arXiv paper out!
"Learning by Surprise": #Surplexity for Mitigation Model Collapse in Generative AI

A new characterization of #modelcollapse in #ai #autophagy and a mitigation inspired by #FreeEnergyPrinciple from cognitive science

@css-conference.bsky.social

arxiv.org/abs/2410.12341
September 3, 2025 at 10:22 AM
dbmutton.bsky.social
A fascinating episode from a superb podcast. It’s the first I’ve heard the phrase “model collapse” and the potential implications for generative AI. Definitely worth a listen if you are actively working with AI #AI #ModelCollapse #GenerativeAI

www.bbc.co.uk/sounds/play/...
The Artificial Human - Will AI Eat Itself? - BBC Sounds
Aleks Krotoski and Kevin Fong uncover what happens when you feed an AI its own output.
www.bbc.co.uk
January 26, 2025 at 11:50 AM
willsh.bsky.social
If this is a topic of interest, come join Prof Alex Bentley next week at lunchtime on Friday over at Newspeak House:

lu.ma/dvbig2kb?tk=...

#modelcollapse
May 30, 2025 at 4:25 PM
paninid.mastodon.world.ap.brid.gy
Whole new meaning to the impact of #modelcollapse
https://tomkahe.com/@GiftArticles/114857402911829126
Gift Articles (@GiftArticles@tomkahe.com)
The government wants AI to fight wars and review your taxes https://wapo.st/4kI5CeC
tomkahe.com
July 15, 2025 at 1:11 PM
mariolo.bsky.social
There is a lot of talk on #modelcollapse of #ai.. it's a fascinating thought that ai may get rid of humans just to then slowly become numb in a kind of incestial process
November 26, 2023 at 9:31 PM
getnews-me.bsky.social
Truncated Cross Entropy (TCE) cuts confidence in AI training, extending performance over two‑fold; fidelity interval 2.3× longer than standard cross‑entropy for text and image models. Read more: https://getnews.me/new-loss-function-delays-model-collapse-in-recursive-ai-training/ #tce #modelcollapse
New Loss Function Delays Model Collapse in Recursive AI Training
September 17, 2025 at 2:00 PM
salibi.bsky.social
هل نواجه "تلوّثًا رقميًا" يُهدد مستقبل #الذكاء_الاصطناعي؟
منذ إطلاق #ChatGPT في 2022، يشبّه خبراء الذكاء الاصطناعي ما حدث بانفجار أول قنبلة ذرية!لماذا ؟
👇👇👇
#AI #ModelCollapse #DataQuality #ChatGPT #ArtificialIntelligence #Ethics #TechPolicy

tinyurl.com/5n9xhc6v
الذكاء الاصطناعي في مواجهة التلوث الرقمي: سباق مع الزمن لحماية “البيانات النقية”
الذكاء الاصطناعي يلوّث نفسه: هل نحن على حافة الانهيار الرقمي؟ منذ إطلاق ChatGPT في نوفمبر 2022، دخل العالم مرحلة جديدة من الذكاء الاصطناعي التوليدي. لكن مع كل هذا الحماس، بدأ بعض الباحثين يرفعون را…
tinyurl.com
June 18, 2025 at 11:39 AM
rderbyshire.bsky.social
I read a paper in 2023, "The Curse of Recursion," which talks about how #AI trained on #SyntheticData can lead to forgotten information and cause a #ModelCollapse. It doesn't mention #Hallucinations, but later papers do.

Could #ChatGPT recent hallucination problem be related? 🤔
May 9, 2025 at 1:31 PM
dbcurren.bsky.social
“A July 2024 paper published in Nature explored the consequences of training #AI models on recursively generated data. It showed that “irreversible defects” can lead to “ #modelcollapse” for systems trained in this way … “ theconversation.com/an-83-year-o... #JorgeLuisBorges #Internet
An 83-year-old short story by Borges portends a bleak future for the internet
Borges imagined an endless library that contained every possible permutation of letters. The truth is out there, but it’s embedded among hoards of lies and gibberish.
theconversation.com
November 20, 2024 at 3:48 AM
drimplausible.com
Soylent Culture.

What is it? Whether it's mass #media, new media, or the media consumed by the current crop of generative #AI tools, it is culture that has been fed on itself. We look at the potential for #ModelCollapse with our AI tools. (audio at the blog)
/15
implausi.blog/2024/11/soyl...
Soylent Culture – The Implausi.blog
implausi.blog
November 16, 2024 at 3:14 PM
bluesky.awakari.com
I love how even the car people are noticing and aware of the AI/LLM/GenAI slop machine. https://www. youtube.com/watch?v=gK_vt3xa6x I # DeadInternetTheory # ModelCollapse

| Details | Interest | Feed |
Origin
birdbutt.com
June 4, 2025 at 3:54 AM
kgadams.net
#ubergeekkellysworld #ai #modelCollapse #LLM #generativeAI #poisonedWell
AI may soon be worthless…
I read an article today over on The Register about how AI may be entering its 'model collapse' phase. The basic idea is that more and more of the content that AI ingests to produce its models is, in fact, AI generated. This means that, at its best, AI's responses will decrease in quality. This sounds very bad, and very predictable.
www.kgadams.net
May 28, 2025 at 1:30 PM
jeremybaker.nz
Generative AI and LLM (large language models) are in serious trouble

jeremybaker.nz/2025/05/30/g...

#ai #aihallucinations #aimodelcollapse #modelcollapse #aibullshit
A selection of screenshots of article’s description the errors, bullshit and model collapse of advanced Generative / LLM AI models.
May 29, 2025 at 10:31 PM
getnews-me.bsky.social
Training diffusion models on their own outputs causes collapse, cutting diversity as model memorizes samples. High‑entropy synthetic data improves diversity (arXiv:2509.16499). Read more: https://getnews.me/model-collapse-from-generalization-to-memorization-in-diffusion-ai/ #modelcollapse #entropy
Model Collapse: From Generalization to Memorization in Diffusion AI
September 24, 2025 at 7:45 AM
kathyreid.au
If you're an #AI company scraping the web for new data to train on, that's bad news, because you also scoop up the AI slop. If models are trained on AI slop, they're likely to encounter #ModelCollapse - like a bad photocopy.

3/5
August 29, 2025 at 7:35 AM
paulguinnessy.bsky.social
Training #artificialintelligence (#AI) models on AI-generated text quickly leads to the models churning out garbage. This phenomenon, termed #ModelCollapse, could halt the improvement of #LargeLanguageModels (#LLMs). Exhibit A: Google search results. www.nature.com/articles/d41...
AI models fed AI-generated data quickly spew nonsense
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 lan...
www.nature.com
July 25, 2024 at 1:25 PM
globalvoters.bsky.social
#AI #modelcollapse TechCrunch Minute: Will AI get dumber?: Although AI models have made big breakthroughs in the last few years, there’s now a risk that they could get dumber and even collapse, according to a new research paper published in Nature. The phenomenon is described by researchers as…
TechCrunch Minute: Will AI get dumber?
Although AI models have made big breakthroughs in the last few years, there’s now a risk that they could get dumber and even collapse, according to a new research paper published in Nature. The phenomenon is described by researchers as “model collapse.”…
dlvr.it
July 30, 2024 at 4:01 PM