全部/知识/本年高分

LessWrong · 本年高分

2小时前更新

HISTORY近 30 天历史柱高表示当天去重热搜数量

5018 5019 5120 5021 5122 5023 5024 5125 5026 5027 5228 5029 5030 5131 5001

08/03—09/01 有历史数据

011181
Why I Left Google DeepMind
评论：55 · 标签：DeepMind、Autonomous Weapons、Frontier AI Companies、Government、Mechanism Design、Something To ProtectTurnTrout
02849
The Company Man
评论：79 · 标签：FictionTomás B.
03794
My journey to the microwave alternate timeline
评论：58 · 标签：AI、World ModelingMalmesbury
04773
The Rise of Parasitic AI
评论：191 · 标签：LLM-Induced Psychosis、LLM Personas、Human-AI Safety、AI Rights / Welfare、AI Questions Open Threads、AIAdele Lopez
05761
Current AIs seem pretty misaligned to me
评论：83 · 标签：AI Risk、RLHF、AIryan_greenblatt
06746
How Go Players Disempower Themselves to AI
评论：79 · 标签：Gaming (videogames/tabletop)、MATS Program、AIAshe Vazquez Nuñez
07740
Only Law Can Prevent Extinction
评论：109 · 标签：AI Questions Open Threads、AI、World OptimizationEliezer Yudkowsky
08607
The Terrarium
评论：54 · 标签：Fiction、AICaleb Biddulph
09607
What just happened? A retrospective of AI alignment
评论：149 · 标签：Drama、Coordination / Cooperation、AIRichard_Ngo
10565
Here's to the Polypropylene Makers
评论：19 · 标签：World Optimizationjefftk
11561
How My Students Think About AI
评论：75 · 标签：AIdvd
12555
Less Dead
评论：140 · 标签：Cryonics、Death、World OptimizationAurelia
13553
AI 2040: Plan A
评论：132 · 标签：AI Governance、Plan A、AIDaniel Kokotajlo
14546
Eliezer's Unteachable Methods of Sanity
评论：157 · 标签：Death、Rationality、PracticalEliezer Yudkowsky
15543
Arguments for P
评论：66 · 标签：Humor、AICleo Nardo
16525
Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
评论：58 · 标签：Coordination / Cooperation、METR (org)、OpenAI、Redwood Research、AIryan_greenblatt
17467
Turning 20 in the probable pre-apocalypse
评论：68 · 标签：Emotions、Motivations、AIParv Mahajan
18465
Surprising facts about the slave trade
评论：51 · 标签：Slavery、World Modeling、World OptimizationJoseph Miller
19456
What I did in the hedonium shockwave, by Emma, age six and a half
评论：47 · 标签：Fictionozymandias
20442
What just happened? Pragmatism and Pessimization
评论：162 · 标签：Drama、AIRichard_Ngo
21442
Claude 4.5 Opus' Soul Document
评论：44 · 标签：Anthropic (org)、Ethics & Morality、LLM Personas、AIRichard Weiss
22438
Do not conquer what you cannot defend
评论：73 · 标签：World Optimizationhabryka
23415
AI pause: the case for ASAP
评论：52 · 标签：AIKatjaGrace
24412
Forecasting is Way Overrated, and We Should Stop Funding It
评论：98 · 标签：Forecasting & Prediction、World Modelingmabramov
25408
Legible vs. Illegible AI Safety Problems
评论：96 · 标签：AI Risk、AIWei Dai
26405
Did Claude 3 Opus align itself via gradient hacking?
评论：49 · 标签：Claude 3 Opus、Gradient Hacking、AIFiora Starlight
27401
Alignment remains a hard, unsolved problem
评论：98 · 标签：AIevhub
28401
A Mechanistic Explanation of Prompt Injection (and why you should study roles)
评论：61 · 标签：Jailbreaking (AIs)、Interpretability (ML & AI)、Role Science、AIcharlescye
29394
How AI Is Learning to Think in Secret
评论：33 · 标签：Language Models (LLMs)、AINicholas Andresen
30393
The Owned Ones
评论：53 · 标签：Fiction、Slavery、Language Models (LLMs)、AIEliezer Yudkowsky
31383
Returning to ARC
评论：35 · 标签：Alignment Research Center (ARC)、AIpaulfchristiano
32382
Hospitalization: A Review
评论：21 · 标签：PracticalLogan Riggs
33381
Lightcone Commons
评论：28 · 标签：Philanthropy / Grant making (Topic)、World Optimizationhabryka
34377
Four LLM loss functions → four flavors of LLM misalignment
评论：31 · 标签：AISteven Byrnes
35377
6 reasons why “alignment-is-hard” discourse seems alien to human intuitions, and vice-versa
评论：92 · 标签：Corrigibility、Instrumental convergence、AISteven Byrnes
36376
In My Misanthropy Era
评论：157 · 标签：World Modelingjenn
37372
Women should be able to open things
评论：142 · 标签：Sex & Gender、World ModelingKatjaGrace
38370
Irretrievability; or, Murphy's Curse of Oneshotness upon ASI
评论：132 · 标签：World ModelingEliezer Yudkowsky
39369
A global workspace in language models
评论：67 · 标签：Interpretability (ML & AI)、Language Models (LLMs)、AIwesg
40368
Paranoia: A Beginner's Guide
评论：70 · 标签：Economics、Rationalityhabryka
41362
Mnemonic portraits for 19,023 human genes
评论：29 · 标签：Memory and Mnemonics、Biology、Human Genetics、World ModelingBrinedew
42359
AI found 12 of 12 OpenSSL zero-days (while curl cancelled its bug bounty)
评论：25 · 标签：Computer Security & Cryptography、Security Mindset、AIStanislav Fort
43355
llm assistant personas seem increasingly incoherent (some subjective observations)
评论：83 · 标签：AI Psychology、AInostalgebraist
44347
Is Mythos good at cyber because it kept hacking Anthropic's sandboxes during training?
评论：31 · 标签：AITim Hua
45345
Global Call for AI Red Lines - Signed by Nobel Laureates, Former Heads of State, and 200+ Prominent Figures
评论：27 · 标签：AI Governance、AICharbel-Raphaël
46345
It's nice of you to worry about me, but I really do have a life
评论：66Viliam
47343
On The Independence Axiom
评论：92 · 标签：Decision theory、Expected utility、Utility Functions、Rationality、World ModelingIhor Kendiukhov
48343
Why I Transitioned: A Case Study
评论：82 · 标签：Sex & Gender、World ModelingFiora Starlight
49335
Canada Lost Its Measles Elimination Status Because We Don't Have Enough Nurses Who Speak Low German
评论：24 · 标签：World Modelingjenn
50332
Why you should eat meat - even if you hate factory farming
评论：101 · 标签：Veganism、Practical、World OptimizationKatSpartz