Band­width Rules Every­thing Around Me: Oliver Habryka on OpenPhil and GoodVentures

ElizabethApr 29, 2025, 8:40 PM
79 points
15 comments1 min readLW link
(acesounderglass.com)

The Grand En­cy­clo­pe­dia of Epony­mous Laws

rogersbaconApr 29, 2025, 7:30 PM
27 points
5 comments16 min readLW link
(www.secretorum.life)

Mis­rep­re­sen­ta­tion as a Bar­rier for In­terp (Part I)

Apr 29, 2025, 5:07 PM
113 points
12 comments7 min readLW link

AISN #53: An Open Let­ter At­tempts to Block OpenAI Restructuring

Apr 29, 2025, 4:13 PM
7 points
0 comments4 min readLW link

What could Alphafold 4 look like?

Abhishaike MahajanApr 29, 2025, 3:45 PM
8 points
0 comments1 min readLW link

Sealed Com­pu­ta­tion: Towards Low-Fric­tion Proof of Locality

Paul BricmanApr 29, 2025, 3:26 PM
4 points
0 comments10 min readLW link
(noemaresearch.com)

Dat­ing Roundup #4: An App for That

ZviApr 29, 2025, 1:10 PM
18 points
5 comments16 min readLW link
(thezvi.wordpress.com)

Talk on let­ters to AI (Lon­don)

ukc10014Apr 29, 2025, 9:50 AM
3 points
0 comments1 min readLW link

Me­mory De­cod­ing Jour­nal Club: “Mo­tor learn­ing se­lec­tively strength­ens cor­ti­cal and stri­atal synapses of mo­tor en­gram neu­rons”

Devin WardApr 29, 2025, 2:26 AM
1 point
0 comments1 min readLW link

D&D.Sci Tax Day: Ad­ven­tur­ers and Assess­ments Eval­u­a­tion & Ruleset

aphyerApr 29, 2025, 2:00 AM
28 points
10 comments5 min readLW link

How to Build a Third Place on Focusmate

Parker ConleyApr 28, 2025, 11:46 PM
96 points
10 comments5 min readLW link
(parconley.com)

Meth­ods of defense against AGI manipulation

MarkelKoriApr 28, 2025, 9:03 PM
3 points
0 comments2 min readLW link

China’s Pe­ti­tion Sys­tem: It Looks Like Democ­racy — But It Isn’t

Hu YichaoApr 28, 2025, 8:56 PM
0 points
4 comments2 min readLW link

Fun­da­men­tals of Safe AI (Phase 1) – Ap­pli­ca­tions Open for the Global Co­hort

rajsecretsApr 28, 2025, 8:52 PM
9 points
0 comments2 min readLW link

Pro­ceed­ings of ILIAD: Les­sons and Progress

Apr 28, 2025, 7:04 PM
78 points
5 comments8 min readLW link

GPT-4o Is An Ab­surd Sycophant

ZviApr 28, 2025, 7:00 PM
81 points
7 comments19 min readLW link
(thezvi.wordpress.com)

[Question] What are the best stan­dard­ised, re­peat­able bets?

kaveApr 28, 2025, 6:45 PM
31 points
10 comments1 min readLW link

7+ tractable di­rec­tions in AI control

Apr 28, 2025, 5:12 PM
89 points
1 comment13 min readLW link

“A vic­tory for the nat­u­ral or­der”

Mati_RoyApr 28, 2025, 3:33 PM
11 points
3 comments1 min readLW link
(preservinghope.substack.com)

Why giv­ing work­ers stocks isn’t enough — and what co-ops get right

B JacobsApr 28, 2025, 2:19 PM
7 points
9 comments2 min readLW link
(bobjacobs.substack.com)

Keltham on Be­com­ing more Truth-Oriented

Towards_KeeperhoodApr 28, 2025, 12:58 PM
22 points
2 comments19 min readLW link

Ther­a­pist in the Weights: Risks of Hyper-In­tro­spec­tion in Fu­ture AI Systems

DavidmanheimApr 28, 2025, 6:42 AM
15 points
1 comment5 min readLW link

In Dark­ness They Assembled

Charlie SandersApr 28, 2025, 3:44 AM
2 points
0 comments3 min readLW link

Seek­ing ad­vice on ca­reers in AI Safety

nemApr 27, 2025, 11:59 PM
8 points
2 comments1 min readLW link

Thin Align­ment Can’t Solve Thick Problems

Daan HenselmansApr 27, 2025, 10:42 PM
11 points
2 comments9 min readLW link

The Way You Go Depends A Good Deal On Where You Want To Get: FEP min­i­mizes sur­prise about ac­tions us­ing prefer­ences about the fu­ture as *ev­i­dence*

Christopher KingApr 27, 2025, 9:55 PM
9 points
5 comments5 min readLW link

How peo­ple use LLMs

ElizabethApr 27, 2025, 9:48 PM
81 points
6 comments1 min readLW link
(www.gleech.org)

Луна Лавгуд и Комната Тайн, Часть 6

Apr 27, 2025, 8:26 PM
3 points
0 comments2 min readLW link

Vic­to­ria, BC

Tomás B.Apr 27, 2025, 5:08 PM
10 points
6 comments1 min readLW link

Our Real­ity: A Si­mu­la­tion Run by a Paper­clip Maximizer

Apr 27, 2025, 4:17 PM
21 points
65 comments5 min readLW link

Ques­tions for old LW mem­bers: how have dis­cus­sions about AI changed com­pared to 10+ years ago?

ExpertiumApr 27, 2025, 4:11 PM
11 points
12 comments1 min readLW link

The case for multi-decade AI timelines [Linkpost]

Noosphere89Apr 27, 2025, 3:31 PM
57 points
22 comments1 min readLW link
(epoch.ai)

My Re­search Pro­cess: Key Mind­sets—Truth-Seek­ing, Pri­ori­ti­sa­tion, Mov­ing Fast

Neel NandaApr 27, 2025, 2:38 PM
42 points
0 comments11 min readLW link

I doubt model col­lapse will hap­pen

HrussApr 27, 2025, 2:08 PM
5 points
0 comments1 min readLW link

Pro­pa­ganda-Bot: A Sketch of a Pos­si­ble RSI

TristanTrimApr 27, 2025, 12:15 PM
6 points
0 comments3 min readLW link

After In­ter­net Dependency

VorakApr 27, 2025, 8:18 AM
14 points
2 comments1 min readLW link

Emer­gence of su­per­in­tel­li­gence from AI hive­minds: how to make it hu­man-friendly?

Mitchell_PorterApr 27, 2025, 4:51 AM
12 points
0 comments2 min readLW link

“The Ur­gency of In­ter­pretabil­ity” (Dario Amodei)

RobertMApr 27, 2025, 4:31 AM
31 points
23 comments3 min readLW link
(www.darioamodei.com)

AI Self Por­traits Aren’t Accurate

JustisMillsApr 27, 2025, 3:27 AM
58 points
10 comments5 min readLW link

MiCARwave

jefftkApr 27, 2025, 2:30 AM
13 points
0 comments1 min readLW link
(www.jefftk.com)

Open Source LLM Poké­mon Scaffold

Julian BradshawApr 27, 2025, 12:57 AM
24 points
0 comments1 min readLW link
(github.com)

What are im­por­tant UI-shaped prob­lems that Light­cone could tackle?

RaemonApr 27, 2025, 12:02 AM
59 points
22 comments2 min readLW link

Kodo and Din

ScrewtapeApr 26, 2025, 6:54 PM
6 points
10 comments4 min readLW link

We should try to au­to­mate AI safety work asap

Marius HobbhahnApr 26, 2025, 4:35 PM
113 points
10 comments15 min readLW link

Re­search Tax­on­omy Gen­er­a­tor and Visualizer

Myles HApr 26, 2025, 4:14 PM
6 points
0 comments6 min readLW link

AI Safety & En­trepreneur­ship v1.0

Chris_LeongApr 26, 2025, 2:37 PM
16 points
0 comments2 min readLW link

Re­con­sid­er­ing Money: The Case for Freigeld in the Digi­tal Age and a Net­worked Future

henophiliaApr 26, 2025, 12:54 PM
−22 points
0 comments5 min readLW link
(blog.hermesloom.org)

How I Think About My Re­search Pro­cess: Ex­plore, Un­der­stand, Distill

Neel NandaApr 26, 2025, 10:31 AM
56 points
5 comments8 min readLW link

Don’t you mean “the most *con­di­tion­ally* for­bid­den tech­nique?”

Knight LeeApr 26, 2025, 3:45 AM
13 points
0 comments3 min readLW link

Land with no aunties

thellimistApr 26, 2025, 1:20 AM
6 points
0 comments1 min readLW link
(kanyilmaz.me)