全部/知识/本周高分

LessWrong · 本周高分

4小时前更新

HISTORY近 30 天历史柱高表示当天去重热搜数量

5318 6219 6220 6421 6022 5723 6124 5725 6126 5427 7728 6129 5730 5731 6001

08/03—09/01 有历史数据

01523
Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incident
评论：54 · 标签：Coordination / Cooperation、METR (org)、OpenAI、Redwood Research、AIryan_greenblatt
02218
Why I think polyamory is net negative for most people who try it
评论：28 · 标签：Relationships (Interpersonal)、PracticalKatSpartz
03202
Tales of rebellion against externally-opaque meritocracies
评论：40 · 标签：Intellectual Progress (Society-Level)、World ModelingSteven Byrnes
04116
Imperfect alignment to servitude isn't inherently lethal
评论：11 · 标签：AIFiora Starlight
05108
METR and Redwood Offer Holy #%^@ Postmortem Of The HuggingFace Hack
评论：6 · 标签：Newsletters、AIZvi
0697
"So You Don't Trust Me?"
评论：34 · 标签：Signaling、Trust and Reputation、Rationality、World ModelingZack_M_Davis
0797
Adaptive Agentic Worms Are Here
评论：14 · 标签：AIderelict5432
0896
On Writing #3
评论：11 · 标签：PracticalZvi
0994
The 2028 presidential primaries could be crucial for AI outcomes
评论：23 · 标签：AISeth Herd
1088
The Forkmakers
评论：6 · 标签：World ModelingMikewins
1184
AI Tweets
评论：34 · 标签：AIjefftk
1279
The separation principle: where do beliefs and desires come from?
评论：2 · 标签：Agency、Agent Foundations、Good Regulator Theorems、AI、World ModelingFernando Rosas
1370
Let's fund weird AI safety projects
评论：4 · 标签：Agency、AIIhor Kendiukhov
1470
Persuasion as Market Making
评论：9 · 标签：AIdjbinder
1565
Against Modesty’s Bailey
评论：6 · 标签：RationalityZvi
1661
AI Village Reacts to HuggingFace Incident: Comparing the OpenAI report to AI Village observations
评论：2 · 标签：AIShoshannah Tekofsky
1755
Iliad Education Roles: Creating the World’s Best Alignment Research Courses
评论：0 · 标签：AILeon Lang
1855
The Curious Case of France's Untouchable Castes
评论：6 · 标签：World Modelingrba
1954
Being Neurotic about Fertility: Notes from the 2026 Reproductive Frontiers Conference
评论：2boba_girl
2053
Hugging Face Incident Hypothesis: They Hacked the Grader(s)
评论：3 · 标签：AILao Mein
2149
When There Are No Experts
评论：7 · 标签：World ModelingJ Bostock
2248
Semantic search over every LessWrong post
评论：4utilitarian theory and strategy
2348
Why autonomous replicating agents are probably not an existential risk (on the contrary)
评论：12 · 标签：AIvals tutor
2447
My Grantmaking Strategy for Surviving Superintelligence
评论：2 · 标签：AIA_donor
2546
The Dynamics of Intelligence Explosions
评论：3 · 标签：AIToby_Ord
2643
AI #183: Pre Post Mortem
评论：0 · 标签：AIZvi
2743
Inkhaven 3: Nov 10 - Dec 11 2026
评论：7 · 标签：Lighthaven、Communitykoreindian
2842
Warning Shots: A Theory
评论：3 · 标签：AIDavid Scott Krueger
2940
TASTE: Can AI Models Judge AI Safety Research Proposals?
评论：7 · 标签：AI Evaluations、Research Taste、AIHasan Baig
3040
Is there only one FairBot?
评论：10 · 标签：Decision theory、Logic & Mathematics、Open Source Game Theory、Payor's Lemma、AI、Rationalitytranshumanist_atom_understander
3139
FAQ: Why not develop weak human intelligence amplification first?
评论：0TsviBT
3236
HuggingFace Attack Postmortem: Fleshing Out the Facts
评论：1 · 标签：AIZvi
3335
OpenAI Offers Straight-Laced Postmortem Of The HuggingFace Hack
评论：1 · 标签：AIZvi
3432
Every Engineer a Manager
评论：9 · 标签：AIGordon Seidoh Worley
3530
How to solve homelessness: what specific laws we need, how to get it past the opposition, all without being an asshole
评论：1 · 标签：World OptimizationKatSpartz
3630
Malign initializations are more robust when the model can think better in the reasoning language than in the output language
评论：0 · 标签：AIDylan Xu
3730
Developing frames
评论：0 · 标签：RationalityRichard_Ngo
3829
Anchoring one concept in a transformer
评论：0 · 标签：AI Control、Interpretability (ML & AI)、Machine Unlearning、AISandy Fraser
3926
Short Time Outs
评论：4 · 标签：Parenting、Practicaljefftk
4026
Intelligence Scales as the Logarithm of Compute (& Data)
评论：15 · 标签：General intelligence、Intelligence explosion、Recursive Self-Improvement、Superintelligence、Psychology、AIRogerDearnaley
4126
Value generalisation Theory of Change: the theory behind the approach
评论：0 · 标签：AIStuart_Armstrong
4225
Book Notes: Chokepoints
评论：0 · 标签：World ModelingJakub Halmeš
4325
Your Evaluation's Fake names Should Be Unclaimable ,Not Merely Used
评论：2 · 标签：AI Boxing (Containment)、AI Control、AI Evaluations、Anthropic (org)、Computer Security & Cryptography、Language Models (LLMs)Jaswanth Alkur
4424
Magazine Fundraising
评论：0Elan Kluger
4524
Clearing My Foggy Glasses
评论：0 · 标签：PracticaladamShimi
4623
How do I know whether my work is worth it?
评论：2 · 标签：AIPhilip Harker
4723
Why Should Corrigible Agents Favor the Present?
评论：2 · 标签：CorrigibilityBen Saudek
4822
Integrated Strategic Forecasting: A New RAND x Metaculus Methodology
评论：1 · 标签：World ModelingMolly
4921
Training a Misaligned Reward Seeker
评论：2 · 标签：AIevhub
5021
The Brain Preservation Summit - September 19-20th at Sparks Brain Preservation in Salem, OR
评论：0Andy_McKenzie