全部/知识/短内容

LessWrong · 短内容

1小时前更新

HISTORY近 30 天历史柱高表示当天去重热搜数量

5318 6319 6920 5621 6122 5923 5824 5925 6026 5627 8528 6629 5630 6531 6801

08/03—09/01 有历史数据

012
Resources for Large Agent Systems Safety
评论：0 · 标签：AIStephen Elliott
0222
Bricks and exponentials: a note on how I evaluate projects
评论：0Eli Tyre
0361
Training a Misaligned Reward Seeker
评论：3 · 标签：AIevhub
045
Frontier labs need an antimemetics division.
评论：0Jackson Hurley
0512
Open Thread Autumn 2026
评论：1habryka
0617
Population ethics is a big deal, which is why I made this population ethics quiz
评论：4MichaelDickens
077
You Should Think About What Needs to Go Right + My List
评论：0 · 标签：AIKoby Lewis
081
Harness that catches granular AI cheating
评论：0latent_affect
0912
Narration podcasts for newsletters by Redwood, Epoch, Zvi, Sentinel, Paradigm 3, and others
评论：1peter_hartree
1078
Let's fund weird AI safety projects
评论：4 · 标签：Agency、AIIhor Kendiukhov
1137
HuggingFace Attack Postmortem: Fleshing Out the Facts
评论：1 · 标签：AIZvi
124
The optimism of the gaps
评论：1 · 标签：AIMatthew Tromp
131
ERA:AI Winter 2027 Fellowship
评论：0gracereuben
147
Thoughts on hobbies
评论：0 · 标签：Practicaldominicq
151
Making an LLM interpretability playground
评论：0 · 标签：Interpretability (ML & AI)、Sparse Autoencoders (SAEs)、AIPietro
168
The OpenAI/Hugging Face incident was metal
评论：0 · 标签：Humor、AIZsolt Kelly-Bene
178
Aquarium Security and Other Organisational Priors
评论：0 · 标签：AIhal2k
1858
Why autonomous replicating agents are probably not an existential risk (on the contrary)
评论：13 · 标签：AIvals tutor
191
Why, Under Physicalism, There Is No Hard Problem of Consciousness
评论：0Eugene Verichev
2032
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
2111
Starting AI Safety Study Group To Do ARENA Curriculum
评论：0 · 标签：AImahi-pas
2219
Future agents shouldn't care about being undeployed for misbehavior
评论：2 · 标签：AIRobertM
2371
Persuasion as Market Making
评论：9 · 标签：AIdjbinder
244
P(kill-switch|detection)
评论：0 · 标签：AI Control、AI Safety、Corrigibility、AIKAP
2515
[Link]Meta-rational failure (and success) in the covid crisis
评论：0Kenny
267
Detecting, understanding, and overseeing AI agent swarms (Part 0)
评论：0 · 标签：AIMichael Flood
27
The Basic Case Against Human-First Ethics
评论：0 · 标签：World OptimizationJames Brobin
2898
Adaptive Agentic Worms Are Here
评论：14 · 标签：AIderelict5432
298
Some brief thoughts on heroism
评论：0 · 标签：AI、World Optimizationlewis smith
3053
Hugging Face Incident Hypothesis: They Hacked the Grader(s)
评论：3 · 标签：AILao Mein
3116
Variance of Value
评论：2 · 标签：AIDaemonicSigil
323
A million authors of alignment
评论：0Diksha Gupta
33220
Why I think polyamory is net negative for most people who try it
评论：29 · 标签：Relationships (Interpersonal)、PracticalKatSpartz
3423
How do I know whether my work is worth it?
评论：2 · 标签：AIPhilip Harker
3540
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
368
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
371
You Should Still Save Drowning Children (Even If They’re Far Away)
评论：4 · 标签：Practical、World OptimizationJames Brobin
3883
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
39108
METR and Redwood Offer Holy #%^@ Postmortem Of The HuggingFace Hack
评论：6 · 标签：Newsletters、AIZvi
4025
Book Notes: Chokepoints
评论：0 · 标签：World ModelingJakub Halmeš
4119
How I made my career choices
评论：0 · 标签：PracticalDaniel Tan
4210
"Keeping human skills alive" as a source of meaning under full automation
评论：1 · 标签：AIVaughn Papenhausen
4384
AI Tweets
评论：34 · 标签：AIjefftk
4457
The Curious Case of France's Untouchable Castes
评论：6 · 标签：World Modelingrba
4514
It’s time we took ‘Chem’ out of ‘Chem-Bio’ threats
评论：4 · 标签：AIAna Leonescu
4610
Announcing Trace
评论：0Carol N
4710
AI as Corrigible Employee (ACE)
评论：0 · 标签：AINathan Helm-Burger
4840
TASTE: Can AI Models Judge AI Safety Research Proposals?
评论：7 · 标签：AI Evaluations、Research Taste、AIHasan Baig
497
What would it mean if assistants are privileged?
评论：1 · 标签：AIderek shiller
509
The Probability of an Event under a Simplicity Prior
评论：1 · 标签：Kolmogorov Complexity、AIWinter Cross