Me­mory De­cod­ing Jour­nal Club: “Sy­nap­tic ar­chi­tec­ture of a mem­ory en­gram in the mouse hip­pocam­pus.”

Devin WardMay 16, 2025, 11:55 PM
3 points
0 comments1 min readLW link

So­cial Anx­iety Isn’t About Be­ing Liked

ChipmonkMay 16, 2025, 10:26 PM
136 points
21 comments2 min readLW link
(chrislakin.blog)

Events: De­bate & Fic­tion Project

abramdemskiMay 16, 2025, 9:51 PM
39 points
1 comment1 min readLW link

How Fast Can Al­gorithms Ad­vance Ca­pa­bil­ities? | Epoch Gra­di­ent Update

henryjMay 16, 2025, 9:38 PM
37 points
8 comments6 min readLW link
(epoch.ai)

P-Values Know When You’re Cheating

EggsMay 16, 2025, 8:34 PM
21 points
2 comments2 min readLW link

Minds are magic

k64May 16, 2025, 7:10 PM
0 points
1 comment2 min readLW link

US-China trade talks should pave way for AI safety treaty [SCMP cross­post]

otto.bartenMay 16, 2025, 4:55 PM
10 points
0 comments3 min readLW link

Direct Real­ism is prob­a­bly false

TerriLeafMay 16, 2025, 4:36 PM
−3 points
19 comments3 min readLW link

Re­gard­ing South Africa

ZviMay 16, 2025, 4:10 PM
71 points
5 comments11 min readLW link
(thezvi.wordpress.com)

Notes on Consciousness

CSDDMay 16, 2025, 2:17 PM
3 points
3 comments1 min readLW link

re­flect­ing on criticism

Vadim GolubMay 16, 2025, 11:59 AM
4 points
5 comments10 min readLW link

Gen­er­at­ing the Fun­niest Joke with RL (ac­cord­ing to GPT-4.1)

aggMay 16, 2025, 5:09 AM
99 points
22 comments4 min readLW link

In­ter­pretable Fine Tun­ing Re­search Up­date and Work­ing Prototype

Matthew KhoriatyMay 16, 2025, 3:44 AM
9 points
0 comments4 min readLW link

It Is Un­ten­able That Near-Fu­ture AI Sce­nario Models Like “AI 2027” Don’t In­clude Open Source AI

Andrew DicksonMay 16, 2025, 2:20 AM
36 points
17 comments5 min readLW link

Ap­ply to Visit­ing Fel­lows at Con­stel­la­tion, due June 13

Ella MarkianosMay 16, 2025, 2:20 AM
1 point
0 comments2 min readLW link

Para­noid Debating

DresdenHeartMay 16, 2025, 2:20 AM
1 point
0 comments1 min readLW link

Bay Area Sum­mer Solstice

May 16, 2025, 12:20 AM
20 points
0 comments1 min readLW link

Stay­ing in a Cap­sule Hotel

jefftkMay 16, 2025, 12:20 AM
19 points
2 comments1 min readLW link
(www.jefftk.com)

Re­search­ing Syn­thetic Con­scious­ness: sound ap­peal­ing?

Brad DunnMay 15, 2025, 10:29 PM
10 points
1 comment1 min readLW link

Start­ing Over: What to tell Sarah, at the edge of pro­fes­sional oblivion.

Brad DunnMay 15, 2025, 9:34 PM
11 points
1 comment20 min readLW link

Tax-Op­ti­mized Risk in Port­fo­lio Allocation

Brendan LongMay 15, 2025, 6:53 PM
6 points
0 comments1 min readLW link
(www.brendanlong.com)

AI Safety Thurs­days: Un­der­stand­ing The Self-Other Over­lap Approach

Juliana EberschlagMay 15, 2025, 6:41 PM
2 points
0 comments1 min readLW link

Some skep­ti­cism about skep­ti­cism about effi­cacy of paus­ing AI

extinction-bountiesMay 15, 2025, 6:15 PM
5 points
1 comment2 min readLW link

time is event based

thiccythotMay 15, 2025, 6:07 PM
45 points
1 comment4 min readLW link

Con­sider Others’ Cost Tolerances

nomagicpillMay 15, 2025, 5:43 PM
23 points
2 comments4 min readLW link
(nomagicpill.github.io)

Prob­lems with in­struc­tion-fol­low­ing as an al­ign­ment target

Seth HerdMay 15, 2025, 3:41 PM
48 points
14 comments10 min readLW link

AI #116: If Any­one Builds It, Every­one Dies

ZviMay 15, 2025, 3:10 PM
47 points
5 comments42 min readLW link
(thezvi.wordpress.com)

Counter-con­sid­er­a­tions on AI arms races

May 15, 2025, 2:54 PM
22 points
0 comments18 min readLW link

AlphaEvolve

mannatvjainMay 15, 2025, 2:14 PM
29 points
0 comments5 min readLW link
(deepmind.google)

From Com­ments on Ac­countabil­ity Sinks

Martin SustrikMay 15, 2025, 10:20 AM
15 points
2 comments7 min readLW link
(250bpm.substack.com)

What Does It Mean to “Write Like You Talk”?

Arjun PanicksseryMay 15, 2025, 9:49 AM
68 points
8 comments5 min readLW link
(arjunpanickssery.substack.com)

What if Agent-4 breaks out?

Alvin ÅnestrandMay 15, 2025, 9:15 AM
10 points
0 comments6 min readLW link

Me­mory De­cod­ing Jour­nal Club: Sy­nap­tic ar­chi­tec­ture of a mem­ory en­gram in the mouse hip­pocam­pus

Devin WardMay 15, 2025, 4:14 AM
1 point
0 comments1 min readLW link

[Question] Why OpenAI pro­jects only $174B of rev­enue by 2030?

becausecuriousMay 15, 2025, 2:50 AM
28 points
6 comments1 min readLW link

Elas­tomeric Fit­ting Session

jefftkMay 15, 2025, 1:50 AM
13 points
4 comments2 min readLW link
(www.jefftk.com)

Re SMTM: nega­tive feed­back on nega­tive feedback

Steven ByrnesMay 14, 2025, 7:50 PM
46 points
1 comment22 min readLW link

Cu­rate your space

Logan KiellerMay 14, 2025, 7:35 PM
21 points
0 comments3 min readLW link
(agenticconjectures.substack.com)

Eliezer and I wrote a book: If Any­one Builds It, Every­one Dies

So8resMay 14, 2025, 7:00 PM
609 points
108 comments2 min readLW link

Notes on Life

CSDDMay 14, 2025, 6:46 PM
−1 points
0 comments5 min readLW link

Xi Jin­ping’s read­out af­ter an AI “study ses­sion” [Chi­naTalk Linkpost]

SodiumMay 14, 2025, 6:25 PM
27 points
1 comment1 min readLW link

Mo­ral Obli­ga­tion and Mo­ral Opportunity

Alice BlairMay 14, 2025, 4:42 PM
37 points
7 comments3 min readLW link

Dodg­ing sys­tem­atic hu­man er­rors in scal­able oversight

Geoffrey IrvingMay 14, 2025, 3:19 PM
33 points
3 comments4 min readLW link

Fight­ing Ob­vi­ous Non­sense About AI Diffusion

ZviMay 14, 2025, 2:20 PM
41 points
8 comments21 min readLW link
(thezvi.wordpress.com)

Achieve­ment Un­locked − 2E Ra­tion­al­ist School in Brooklyn

Laura BMay 14, 2025, 1:02 PM
27 points
0 comments3 min readLW link

Ex­am­in­ing Batenka’s Le­gal Per­son­hood Frame­work for Models

Stephen MartinMay 14, 2025, 10:15 AM
5 points
4 comments7 min readLW link

Doomers Should try Much Harder to get Famous

HideMay 14, 2025, 6:15 AM
13 points
4 comments3 min readLW link

De­fault ar­gu­ments in ca­sual speech

Adam ZernerMay 14, 2025, 4:00 AM
15 points
0 comments4 min readLW link

The Best Refer­ence Works for Every Subject

Parker ConleyMay 14, 2025, 12:58 AM
124 points
22 comments6 min readLW link
(parconley.com)

Oc­to­ber The First Is Too Late

gwernMay 13, 2025, 9:45 PM
58 points
8 comments1 min readLW link
(gwern.net)

An­nounc­ing Tra­jec­tory Labs—A Toronto AI Safety Office

May 13, 2025, 9:04 PM
24 points
3 comments2 min readLW link
(forum.effectivealtruism.org)