Still yet one more thing people have been predicting is happening in the #ai world: model collapse from consumption of ai-generated web content.
“Recursive pollution is a serious threat to LLM integrity. ML systems should not eat their own output just as mammals should not consume brains of their own species.”
— Berryville Institute of ML, ARA of LLMs
https://www.theregister.com/2025/05/27/opinion_column_ai_model_collapse
Brian Sletten
npub1jg…k45wa
2025-05-30 15:22:17 UTC
Author Public Key
npub1jglgftjhjqmh0fy7tvr655mtretyu4r5tv3znr34z08a54d6wlsslk45waSeen on
wss://relay.momostr.pinkPublished at
2025-05-30 15:22:17 UTCEvent JSON
{
"id": "2d5fac1adedce7211dd73d2e2bc43a9a9bfd2bccae23b28c34e4bdf28862fc54",
"pubkey": "923e84ae57903777a49e5b07aa536b1e564e54745b22298e3513cfda55ba77e1",
"created_at": 1748618537,
"kind": 1,
"tags": [
[
"proxy",
"https://mastodon.social/@bsletten/114597464446199899",
"web"
],
[
"t",
"ai"
],
[
"proxy",
"https://mastodon.social/users/bsletten/statuses/114597464446199899",
"activitypub"
],
[
"L",
"pink.momostr"
],
[
"l",
"pink.momostr.activitypub:https://mastodon.social/users/bsletten/statuses/114597464446199899",
"pink.momostr"
],
[
"-"
]
],
"content": "Still yet one more thing people have been predicting is happening in the #ai world: model collapse from consumption of ai-generated web content.\n\n“Recursive pollution is a serious threat to LLM integrity. ML systems should not eat their own output just as mammals should not consume brains of their own species.”\n\n— Berryville Institute of ML, ARA of LLMs\n\nhttps://www.theregister.com/2025/05/27/opinion_column_ai_model_collapse",
"sig": "3f57fea852082de0c554599a4db5bd401297f446af5b3053fc14548ab76ef250f1d80d4922c062886aff2e9598560b78365e2df835a52f6b88f75ac21373a8c7"
}