Why Can’t We Hy­poth­e­size After the Fact?

David UdellFeb 26, 2025, 10:41 PM
40 points
3 comments2 min readLW link

“AI Rapidly Gets Smarter, And Makes Some of Us Dum­ber,” from Sabine Hossenfelder

Evan_GaensbauerFeb 26, 2025, 10:33 PM
4 points
9 comments2 min readLW link
(youtu.be)

METR: AI mod­els can be dan­ger­ous be­fore pub­lic deployment

UnofficialLinkpostBotFeb 26, 2025, 8:19 PM
16 points
0 comments3 min readLW link
(metr.org)

Rep­re­sen­ta­tion Eng­ineer­ing has Its Prob­lems, but None Seem Unsolvable

Lukasz G BartoszczeFeb 26, 2025, 7:53 PM
15 points
1 comment3 min readLW link

Thoughts that prompt good fore­casts: A survey

Daniel_FriedrichFeb 26, 2025, 6:36 PM
1 point
0 comments1 min readLW link

The non-tribal tribes

PatrickDFarleyFeb 26, 2025, 5:22 PM
24 points
4 comments16 min readLW link

SAE Train­ing Dataset In­fluence in Fea­ture Match­ing and a Hy­poth­e­sis on Po­si­tion Features

Seonglae ChoFeb 26, 2025, 5:05 PM
4 points
3 comments17 min readLW link

Fuzzing LLMs some­times makes them re­veal their secrets

Fabien RogerFeb 26, 2025, 4:48 PM
61 points
13 comments9 min readLW link

You can just wear a suit

lsusrFeb 26, 2025, 2:57 PM
111 points
48 comments2 min readLW link

Matthew Ygle­sias—Mis­in­for­ma­tion Mostly Con­fuses Your Own Side

SiebeFeb 26, 2025, 2:55 PM
10 points
1 comment1 min readLW link
(www.slowboring.com)

Op­ti­miz­ing Feed­back to Learn Faster

Towards_KeeperhoodFeb 26, 2025, 2:24 PM
12 points
0 comments2 min readLW link

out­lin­ing is a his­tor­i­cally re­cent un­der­uti­lized gift to family

daijinFeb 26, 2025, 1:58 PM
4 points
2 comments3 min readLW link

Osaka

lsusrFeb 26, 2025, 1:50 PM
73 points
11 comments1 min readLW link

Time to Wel­come Claude 3.7

ZviFeb 26, 2025, 1:00 PM
49 points
2 comments24 min readLW link
(thezvi.wordpress.com)

[PAPER] Ja­co­bian Sparse Au­toen­coders: Spar­sify Com­pu­ta­tions, Not Just Activations

Lucy FarnikFeb 26, 2025, 12:50 PM
79 points
8 comments7 min readLW link

Minor in­ter­pretabil­ity ex­plo­ra­tion #1: Grokking of mod­u­lar ad­di­tion, sub­trac­tion, mul­ti­pli­ca­tion, for differ­ent ac­ti­va­tion functions

Rareș BaronFeb 26, 2025, 11:35 AM
3 points
13 comments4 min readLW link

[Question] Name for Stan­dard AI Caveat?

yrimonFeb 26, 2025, 7:07 AM
6 points
5 comments1 min readLW link

Levels of anal­y­sis for think­ing about agency

Cole WyethFeb 26, 2025, 4:24 AM
11 points
0 comments7 min readLW link

The Stag Hunt—cul­ti­vat­ing co­op­er­a­tion to reap rewards

James Stephen BrownFeb 25, 2025, 11:45 PM
7 points
0 comments4 min readLW link
(nonzerosum.games)

Three Levels for Large Lan­guage Model Cognition

Eleni AngelouFeb 25, 2025, 11:14 PM
21 points
0 comments5 min readLW link

[Cross­post] Strate­gic wealth ac­cu­mu­la­tion un­der trans­for­ma­tive AI expectations

Feb 25, 2025, 9:50 PM
5 points
0 comments17 min readLW link
(forum.effectivealtruism.org)

Emer­gent Misal­ign­ment: Nar­row fine­tun­ing can pro­duce broadly mis­al­igned LLMs

Feb 25, 2025, 5:39 PM
329 points
91 comments4 min readLW link

We Can Build Com­pas­sion­ate AI

Gordon Seidoh WorleyFeb 25, 2025, 4:37 PM
9 points
6 comments4 min readLW link
(uncertainupdates.substack.com)

[Question] In­tel­lec­tual life­hacks repo

Antoine de ScorrailleFeb 25, 2025, 4:32 PM
11 points
15 comments1 min readLW link

Eco­nomics Roundup #5

ZviFeb 25, 2025, 1:40 PM
27 points
10 comments20 min readLW link
(thezvi.wordpress.com)

Mak­ing al­ign­ment a law of the universe

Richard JugginsFeb 25, 2025, 10:44 AM
0 points
3 comments15 min readLW link

Re­vis­it­ing Con­way’s Law

annebrandesFeb 25, 2025, 8:33 AM
12 points
4 comments3 min readLW link

De­mys­tify­ing the Pinoc­chio Paradox

Novak ZukowskiFeb 25, 2025, 6:16 AM
−1 points
0 comments3 min readLW link

Tech­ni­cal com­par­i­son of Deepseek, No­vasky, S1, Helix, P0

JuliezhangggFeb 25, 2025, 4:20 AM
8 points
0 comments5 min readLW link

Up­com­ing Protest for AI Safety

Matt VincentFeb 25, 2025, 3:04 AM
12 points
0 comments1 min readLW link
(www.pauseai-us.org)

what an effi­cient mar­ket feels from inside

DMMFFeb 25, 2025, 2:38 AM
40 points
9 comments6 min readLW link
(danfrank.ca)

Metacompilation

Donald HobsonFeb 24, 2025, 10:58 PM
11 points
1 comment4 min readLW link

The man­i­fest manifesto

dkl9Feb 24, 2025, 10:13 PM
6 points
2 comments2 min readLW link
(dkl9.net)

Credit Suisse col­lapse obfus­cated Par­reaux, Thiébaud & Part­ners scan­dal

pocockFeb 24, 2025, 9:28 PM
3 points
0 comments1 min readLW link
(juristgate.com)

Topolog­i­cal Data Anal­y­sis and Mechanis­tic Interpretability

Gunnar CarlssonFeb 24, 2025, 7:56 PM
16 points
4 comments7 min readLW link

Zizian com­par­i­sons /​ con­nec­tions in the open source & Linux communities

pocockFeb 24, 2025, 7:55 PM
−15 points
0 comments1 min readLW link

Lo­cal Trust

Feb 24, 2025, 7:53 PM
21 points
4 comments5 min readLW link

Na­tion­wide Ac­tion Work­shop: Con­tact Congress about AI safety!

Felix De SimoneFeb 24, 2025, 7:36 PM
7 points
0 comments1 min readLW link

An­thropic re­leases Claude 3.7 Son­net with ex­tended think­ing mode

LawrenceCFeb 24, 2025, 7:32 PM
88 points
8 comments4 min readLW link
(www.anthropic.com)

Train­ing AI to do al­ign­ment re­search we don’t already know how to do

joshcFeb 24, 2025, 7:19 PM
45 points
23 comments7 min readLW link

Con­fer­ence Re­port: Thresh­old 2030 - Model­ing AI Eco­nomic Futures

Feb 24, 2025, 6:56 PM
51 points
0 comments10 min readLW link
(www.convergenceanalysis.org)

Eval­u­at­ing “What 2026 Looks Like” So Far

Jonny SpicerFeb 24, 2025, 6:55 PM
77 points
5 comments7 min readLW link

Su­per­in­tel­li­gent Agents Pose Catas­trophic Risks: Can Scien­tist AI Offer a Safer Path?

Feb 24, 2025, 6:31 PM
44 points
15 comments11 min readLW link

Un­der­stand­ing Agent Preferences

martinkunevFeb 24, 2025, 5:46 PM
6 points
2 comments14 min readLW link

What We Can Do to Prevent Ex­tinc­tion by AI

Joe RogeroFeb 24, 2025, 5:15 PM
12 points
0 commentsLW link

Dream, Truth, & Good

abramdemskiFeb 24, 2025, 4:59 PM
50 points
11 comments4 min readLW link

Fore­cast­ing Fron­tier Lan­guage Model Agent Capabilities

Feb 24, 2025, 4:51 PM
35 points
0 comments5 min readLW link
(www.apolloresearch.ai)

A City Within a City

Declan MolonyFeb 24, 2025, 3:51 PM
48 points
1 comment7 min readLW link

Grok Grok

ZviFeb 24, 2025, 2:20 PM
36 points
2 comments19 min readLW link
(thezvi.wordpress.com)

if you’re not happy sin­gle, you won’t be happy immortal

daijinFeb 24, 2025, 1:23 PM
2 points
1 comment1 min readLW link