pull down to refresh

AI models collapse when trained on recursively generated data - Naturewww.nature.com/articles/s41586-024-07566-y
52 sats \ 0 comments \ @Rsync25 24 Jul 2024 alter_native
related
Less is More: Recursive Reasoning w/ Tiny Networks - Alexia Jolicoeur-Martineaugithub.com/SamsungSAILMontreal/TinyRecursiveModels
338 sats \ 1 comment \ @Scoresby 8 Oct 2025 AI
Recursive Language Modelsalexzhang13.github.io/blog/2025/rlm/
187 sats \ 0 comments \ @carter 16 Oct 2025 AI
Elites, the curse of recursion, and the half-life of policy
5779 sats \ 11 comments \ @elvismercury 29 Mar 2024 mostly_harmless
Less is More: Recursive Reasoning with Tiny Networksarxiv.org/html/2510.04871v1
65 sats \ 1 comment \ @byzantine 9 Oct 2025 AI
In a First, AI Models Analyze Language As Well As a Human Expertwww.quantamagazine.org/in-a-first-ai-models-analyze-language-as-well-as-a-human-expert-20251031/
274 sats \ 0 comments \ @0xbitcoiner 31 Oct 2025 AI
AI model collapse is not what we paid forwww.theregister.com/2025/05/27/opinion_column_ai_model_collapse/
451 sats \ 0 comments \ @k00b 28 May 2025 tech
AI devs close to scraping bottom of data barrelwww.theregister.com/2025/10/03/ai_training_requires_more_data/
457 sats \ 10 comments \ @0xbitcoiner 5 Oct 2025 AI
Verbalized Sampling: How to Mitigate Mode Collapse and Unlock LLM Diversityarxiv.org/abs/2510.01171
196 sats \ 0 comments \ @Scoresby 17 Oct 2025 AI
Has generative AI already peaked?youtu.be/dDUC-LqVrPU
1368 sats \ 19 comments \ @davidw 10 May 2024 AI freebie
The AI Industry's Scaling Obsession Is Headed for a Cliffarxiv.org/abs/2507.07931
351 sats \ 9 comments \ @0xbitcoiner 15 Oct 2025 AI
Distinct AI Models Seem To Converge On How They Encode Realitywww.quantamagazine.org/distinct-ai-models-seem-to-converge-on-how-they-encode-reality-20260107/
197 sats \ 1 comment \ @0xbitcoiner 7 Jan AI
Is It EVEN Possible To Reverse Engineer AI’s Training Data?www.youtube.com/watch?v=Nie5L7Z6-qk
199 sats \ 0 comments \ @carter 21 Sep 2025 AI
Oxford Researcher Warns That AI Is Heading for a Hindenburg-Style Disasterfuturism.com/artificial-intelligence/ai-hindenburg-disast
160 sats \ 0 comments \ @winteryeti 19 Feb AI
As human generated data is used up, companies resort to "Synthetic data"www.nytimes.com/2024/04/06/technology/tech-giants-harvest-data-artificial-intelligence.html?ugrp=u&unlocked_article_code=1.i00.2gRB.J2qpf6PYyX6b&smid=url-share
3518 sats \ 7 comments \ @chaoticalHeavy 8 Apr 2024 AI
Semantic ablation: Why AI writing is boring and dangerouswww.theregister.com/2026/02/16/semantic_ablation_ai_writing/
321 sats \ 2 comments \ @0xbitcoiner 17 Feb AI
Researchers isolate memorization from problem-solving in AI neural networksarstechnica.com/ai/2025/11/study-finds-ai-models-store-memories-and-logic-in-different-neural-regions/
430 sats \ 1 comment \ @0xbitcoiner 11 Nov 2025 AI
AI training data set size is growing at 260% per year, compute at 360% per year
509 sats \ 1 comment \ @k00b 31 May 2025 charts_and_maps
Boffins craft certified way for AI to unlearn private datawww.theregister.com/2025/09/04/boffins_detail_ai_mind_wipe/
252 sats \ 6 comments \ @0xbitcoiner 5 Sep 2025 AI
AI agents can't teach themselves new tricks – people canwww.theregister.com/2026/02/19/ai_agents_cant_teach_themselves/
221 sats \ 1 comment \ @0xbitcoiner 19 Feb AI
Researchers poison stolen data to make AI results wrongwww.theregister.com/2026/01/06/ai_data_pollution_defense/
266 sats \ 1 comment \ @0xbitcoiner 6 Jan AI
I Trained a Custom Model…
2111 sats \ 12 comments \ @antic 30 Dec 2023 art