pull down to refresh

Emergent Misalignment: Narrow finetuning can produce broadly misaligned LLMsarxiv.org/abs/2502.17424
227 sats \ 6 comments \ @carter 14 Jul 2025 AI
related
Moloch's Bargain: Emergent Misalignment When LLMs Compete for Audiencesarxiv.org/abs/2510.06105
231 sats \ 1 comment \ @carter 9 Oct 2025 AI
Weird Generalization and Inductive Backdoors: New Ways to Corrupt LLMsarxiv.org/abs/2512.09742
401 sats \ 2 comments \ @Scoresby 14 Dec 2025 AI
Systemic Misalignmentwww.systemicmisalignment.com/
185 sats \ 0 comments \ @carter 30 Jun 2025 AI
Agentic Misalignment: How LLMs could be insider threatswww.anthropic.com/research/agentic-misalignment
130 sats \ 0 comments \ @carter 8 Aug 2025 AI
Learning the Wrong Lessons: Syntactic-Domain Spurious Correlations in LLMsarxiv.org/abs/2509.21155
445 sats \ 6 comments \ @optimism 2 Dec 2025 AI
2025 LLM Year in Review - karpathykarpathy.bearblog.dev/year-in-review-2025/
1652 sats \ 3 comments \ @Scoresby 21 Dec 2025 AI
The week in AI, October 6-12, 2025
991 sats \ 2 comments \ @optimism 13 Oct 2025 AI
Context Rot: How Increasing Input Tokens Impacts LLM Performanceresearch.trychroma.com/context-rot
334 sats \ 2 comments \ @Scoresby 14 Jul 2025 AI
The simulation of judgment in LLMs - PNASwww.pnas.org/doi/10.1073/pnas.2518443122
244 sats \ 5 comments \ @Scoresby 15 Oct 2025 AI
Elites, the curse of recursion, and the half-life of policy
5779 sats \ 11 comments \ @elvismercury 29 Mar 2024 mostly_harmless
When Models Lie, We Learn: Multilingual Span-Level Hallucination Detectionarxiv.org/abs/2510.04849v1
433 sats \ 2 comments \ @optimism 19 Oct 2025 AI
LLMs Can Get Brain Rotllm-brain-rot.github.io/
287 sats \ 0 comments \ @Scoresby 21 Oct 2025 AI
LLMs generate slop because they avoid surprises by design - Dan Fabulichdanfabulich.medium.com/llms-tell-bad-jokes-because-they-avoid-surprises-7f111aac4f96
373 sats \ 2 comments \ @Scoresby 19 Aug 2025 AI
LLMs: a bleak future ahead?lcamtuf.substack.com/p/llms-a-bleak-future-ahead
266 sats \ 5 comments \ @cointastical 9 Jan 2023 bitcoin
Show HN: 80% faster, 50% less memory, 0% loss of accuracy Llama finetuninggithub.com/unslothai/unsloth
31 sats \ 2 comments \ @hn 2 Dec 2023 tech
More Artificial than Intelligent, it is only getting worse - Mathjis Lagerbergmlagerberg.com/much-a-little-i-and-it-is-not-getting-better/
247 sats \ 4 comments \ @Scoresby 15 Jul 2025 AI
Andrej Karpathy: How I use LLMswww.youtube.com/watch?v=EWvNQjAaOHw
1278 sats \ 1 comment \ @k00b 28 Feb 2025 AI
Why do LLMs have emergent properties?www.johndcook.com/blog/2025/05/08/why-do-llms-have-emergent-properties/
561 sats \ 0 comments \ @k00b 9 May 2025 tech
Detecting and reducing scheming in AI models - OpenAIopenai.com/index/detecting-and-reducing-scheming-in-ai-models/
257 sats \ 1 comment \ @Scoresby 17 Sep 2025 AI
If you think LLMs produce overly defensive code, you're not alone
150 sats \ 0 comments \ @tonyaldon 21 Dec 2025 devs
Things we learned about LLMs in 2024simonwillison.net/2024/Dec/31/llms-in-2024/
470 sats \ 0 comments \ @Rsync25 31 Dec 2024 tech