全部/知识/算法首页

LessWrong · 算法首页

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 6019 5820 6021 6022 5923 5824 5825 5926 5327 7428 6329 6330 6931 6101

08/03—09/01 有历史数据

019
Training a Misaligned Reward Seeker
评论：1 · 标签：AIevhub
0265
Let's fund weird AI safety projects
评论：4 · 标签：Agency、AIIhor Kendiukhov
0312
Population ethics is a big deal, which is why I made this population ethics quiz
评论：1MichaelDickens
0479
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
05218
Why I think polyamory is net negative for most people who try it
评论：27 · 标签：Relationships (Interpersonal)、PracticalKatSpartz
0647
Why autonomous replicating agents are probably not an existential risk (on the contrary)
评论：12 · 标签：AIvals tutor
078
prefilling an emergently misaligned model with reasoning traces that produced misaligned answers increases misalignment rates by ~8% but reasoning traces that produce misaligned answers aren't detectable through text monitoring
评论：0 · 标签：AInesiacel
087
You Should Think About What Needs to Go Right + My List
评论：0 · 标签：AIKoby Lewis
09523
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
10202
Tales of rebellion against externally-opaque meritocracies
评论：40 · 标签：Intellectual Progress (Society-Level)、World ModelingSteven Byrnes
1197
Adaptive Agentic Worms Are Here
评论：14 · 标签：AIderelict5432
1270
Persuasion as Market Making
评论：9 · 标签：AIdjbinder
135
The inverse Simmel scenario
评论：1 · 标签：Rationality、World ModelingFernand0
1417
A Catalogue of Corrigibility Counter Arguments
评论：6 · 标签：Corrigibility、AIFelix Moses
153
The Structuralist Mindset
评论：0 · 标签：World ModelingFernand0
162
Wiki-libertarianism
评论：1 · 标签：World OptimizationFernand0
17442
What just happened? Pragmatism and Pessimization
评论：160 · 标签：Drama、AIRichard_Ngo
1853
Hugging Face Incident Hypothesis: They Hacked the Grader(s)
评论：3 · 标签：AILao Mein
199
Value generalisation Theory of Change: putting it into practice
评论：1 · 标签：AIStuart_Armstrong
207
Thoughts on hobbies
评论：0 · 标签：Practicaldominicq
214
The optimism of the gaps
评论：1 · 标签：AIMatthew Tromp
22
Kantian consent
评论：2 · 标签：World Modeling、World OptimizationFernand0
2384
AI Tweets
评论：34 · 标签：AIjefftk
24116
Imperfect alignment to servitude isn't inherently lethal
评论：11 · 标签：AIFiora Starlight
25245
AI Safety Acculturation is Neglected
评论：55 · 标签：AIjenn
2626
Intelligence Scales as the Logarithm of Compute (& Data)
评论：15 · 标签：General intelligence、Intelligence explosion、Recursive Self-Improvement、Superintelligence、Psychology、AIRogerDearnaley
2719
Future agents shouldn't care about being undeployed for misbehavior
评论：1 · 标签：AIRobertM
28561
How My Students Think About AI
评论：75 · 标签：AIdvd
2940
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
3094
The 2028 presidential primaries could be crucial for AI outcomes
评论：23 · 标签：AISeth Herd
318
Study 2 Results: Exploring representational counterparts of welfare-relevant indicators under post-training quantization
评论：0 · 标签：AIashesfall
3216
AIs Thinking Dangerous Thoughts
评论：5 · 标签：AI、RationalityChantiel
3355
The Curious Case of France's Untouchable Castes
评论：6 · 标签：World Modelingrba
34607
What just happened? A retrospective of AI alignment
评论：149 · 标签：Drama、Coordination / Cooperation、AIRichard_Ngo
3515
[Link]Meta-rational failure (and success) in the covid crisis
评论：0Kenny
3613
OODing up your Anki game
评论：1 · 标签：Practical、RationalityFernand0
3797
"So You Don't Trust Me?"
评论：34 · 标签：Signaling、Trust and Reputation、Rationality、World ModelingZack_M_Davis
381181
Why I Left Google DeepMind
评论：55 · 标签：DeepMind、Autonomous Weapons、Frontier AI Companies、Government、Mechanism Design、Something To ProtectTurnTrout
39126
PSA: We can do better
评论：14 · 标签：AIhersheys
4016
Variance of Value
评论：2 · 标签：AIDaemonicSigil
4161
AI Village Reacts to HuggingFace Incident: Comparing the OpenAI report to AI Village observations
评论：2 · 标签：AIShoshannah Tekofsky
4296
On Writing #3
评论：11 · 标签：PracticalZvi
438
Thoughts I failed to develop into posts
评论：3Fernand0
44228
RL creates split personas
评论：19 · 标签：AIJan Betley
45144
Twenty Years from RSI to Takeoff: Slow Learning, Scaling Slowdown, Industrial Explosion
评论：33 · 标签：AI Timelines、Compute、Language Models (LLMs)、AIVladimir_Nesov
4647
My Grantmaking Strategy for Surviving Superintelligence
评论：2 · 标签：AIA_donor
47543
Arguments for P
评论：66 · 标签：Humor、AICleo Nardo
4817
An Unsimulated Simulation
评论：0 · 标签：World Modelingsil
4940
TASTE: Can AI Models Judge AI Safety Research Proposals?
评论：7 · 标签：AI Evaluations、Research Taste、AIHasan Baig
5046
The Dynamics of Intelligence Explosions
评论：3 · 标签：AIToby_Ord