AI 2027: What Su­per­in­tel­li­gence Looks Like

Apr 3, 2025, 4:23 PM
657 points
222 comments41 min readLW link
(ai-2027.com)

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

So8resMay 14, 2025, 7:00 PM
644 points
114 comments2 min readLW link

How to Make Superbabies

Feb 19, 2025, 8:39 PM
619 points
355 comments31 min readLW link

Ori­ent­ing Toward Wizard Power

johnswentworthMay 8, 2025, 5:23 AM
558 points
146 comments5 min readLW link

A case for courage, when speak­ing of AI danger

So8resJun 27, 2025, 2:15 AM
513 points
128 comments6 min readLW link

New En­dorse­ments for “If Any­one Builds It, Every­one Dies”

MaloJun 18, 2025, 4:30 PM
484 points
55 comments4 min readLW link
(intelligence.org)

What We Learned from Briefing 70+ Law­mak­ers on the Threat from AI

leticiagarciaMay 27, 2025, 6:23 PM
482 points
15 comments16 min readLW link
(substack.com)

Ac­countabil­ity Sinks

Martin SustrikApr 22, 2025, 5:00 AM
432 points
57 comments15 min readLW link
(250bpm.substack.com)

How AI Takeover Might Hap­pen in 2 Years

joshcFeb 7, 2025, 5:10 PM
432 points
139 comments29 min readLW link
(x.com)

How Does A Blind Model See The Earth?

henryAug 11, 2025, 7:58 PM
430 points
38 comments7 min readLW link
(outsidetext.substack.com)

HPMOR: The (Prob­a­bly) Un­told Lore

Jul 25, 2025, 6:39 PM
405 points
149 comments38 min readLW link

Will Je­sus Christ re­turn in an elec­tion year?

Eric NeymanMar 24, 2025, 4:50 PM
401 points
59 comments4 min readLW link
(ericneyman.wordpress.com)

Play­ing in the Creek

HastingsApr 10, 2025, 5:39 PM
387 points
13 comments2 min readLW link
(hgreer.com)

the void

nostalgebraistJun 11, 2025, 3:19 AM
383 points
107 comments1 min readLW link
(nostalgebraist.tumblr.com)

A Bear Case: My Pre­dic­tions Re­gard­ing AI Progress

Thane RuthenisMar 5, 2025, 4:41 PM
372 points
162 comments9 min readLW link

A deep cri­tique of AI 2027’s bad timeline models

titotalJun 19, 2025, 1:29 PM
359 points
39 comments39 min readLW link
(titotal.substack.com)

The Case Against AI Con­trol Research

johnswentworthJan 21, 2025, 4:03 PM
358 points
81 comments6 min readLW link

LessWrong has been ac­quired by EA

habrykaApr 1, 2025, 1:09 PM
354 points
52 comments1 min readLW link

Gen­er­al­ized Han­gri­ness: A Stan­dard Ra­tion­al­ist Stance Toward Emotions

johnswentworthJul 10, 2025, 6:22 PM
353 points
69 comments7 min readLW link

What’s the short timeline plan?

Marius HobbhahnJan 2, 2025, 2:59 PM
353 points
49 comments23 min readLW link

Re­cent AI model progress feels mostly like bullshit

lcMar 24, 2025, 7:28 PM
352 points
85 comments8 min readLW link
(zeropath.com)

VDT: a solu­tion to de­ci­sion theory

L Rudolf LApr 1, 2025, 9:04 PM
342 points
31 comments4 min readLW link

Policy for LLM Writ­ing on LessWrong

jimrandomhMar 24, 2025, 9:41 PM
333 points
70 comments2 min readLW link

Sublimi­nal Learn­ing: LLMs Trans­mit Be­hav­ioral Traits via Hid­den Sig­nals in Data

Jul 22, 2025, 4:37 PM
332 points
33 comments4 min readLW link

Emer­gent Misal­ign­ment: Nar­row fine­tun­ing can pro­duce broadly mis­al­igned LLMs

Feb 25, 2025, 5:39 PM
330 points
92 comments4 min readLW link

In­ter­pretabil­ity Will Not Reli­ably Find De­cep­tive AI

Neel NandaMay 4, 2025, 4:32 PM
321 points
68 comments7 min readLW link

Mur­der plots are infohazards

Chris MonteiroFeb 13, 2025, 7:15 PM
308 points
44 comments2 min readLW link

Trac­ing the Thoughts of a Large Lan­guage Model

Adam JermynMar 27, 2025, 5:20 PM
305 points
24 comments10 min readLW link
(www.anthropic.com)

So You Think You’ve Awo­ken ChatGPT

JustisMillsJul 11, 2025, 1:01 AM
299 points
85 comments9 min readLW link

The Problem

Aug 5, 2025, 9:40 PM
298 points
211 comments26 min readLW link

So You Want To Make Marginal Progress...

johnswentworthFeb 7, 2025, 11:22 PM
296 points
42 comments4 min readLW link

Make More Grayspaces

Duncan Sabien (Inactive)Jul 19, 2025, 10:22 PM
293 points
65 comments13 min readLW link

Be­ware Gen­eral Claims about “Gen­er­al­iz­able Rea­son­ing Ca­pa­bil­ities” (of Modern AI Sys­tems)

LawrenceCJun 11, 2025, 7:27 PM
293 points
19 comments16 min readLW link

Good Re­search Takes are Not Suffi­cient for Good Strate­gic Takes

Neel NandaMar 22, 2025, 10:13 AM
292 points
28 comments4 min readLW link
(www.neelnanda.io)

Ar­bital has been im­ported to LessWrong

Feb 20, 2025, 12:47 AM
281 points
30 comments5 min readLW link

Why Have Sen­tence Lengths De­creased?

Arjun PanicksseryApr 3, 2025, 5:50 PM
280 points
89 comments4 min readLW link
(arjunpanickssery.substack.com)

Foom & Doom 1: “Brain in a box in a base­ment”

Steven ByrnesJun 23, 2025, 5:18 PM
276 points
120 comments29 min readLW link

Truth or Dare

Duncan Sabien (Inactive)May 29, 2025, 12:07 AM
272 points
58 comments69 min readLW link

Love stays loved (formerly “Skin”)

Swimmer963 (Miranda Dixon-Luinenburg) Jul 18, 2025, 7:17 PM
267 points
12 comments29 min readLW link

the jack­pot age

thiccythotJul 11, 2025, 9:05 PM
257 points
17 comments4 min readLW link

Why Should I As­sume CCP AGI is Worse Than USG AGI?

Tomás B.Apr 19, 2025, 2:47 PM
255 points
87 comments1 min readLW link

Tro­jan Sky

Richard_NgoMar 11, 2025, 3:14 AM
250 points
39 comments12 min readLW link
(www.narrativeark.xyz)

The Gen­tle Romance

Richard_NgoJan 19, 2025, 6:29 PM
243 points
46 comments15 min readLW link
(www.asimov.press)

METR: Mea­sur­ing AI Abil­ity to Com­plete Long Tasks

Zach Stein-PerlmanMar 19, 2025, 4:00 PM
241 points
106 comments5 min readLW link
(metr.org)

To Un­der­stand His­tory, Keep Former Pop­u­la­tion Distri­bu­tions In Mind

Arjun PanicksseryApr 23, 2025, 4:51 AM
240 points
13 comments2 min readLW link
(arjunpanickssery.substack.com)

A His­tory of the Fu­ture, 2025-2040

L Rudolf LFeb 17, 2025, 12:03 PM
237 points
42 comments75 min readLW link
(nosetgauge.substack.com)

Four ways learn­ing Econ makes peo­ple dumber re: fu­ture AI

Steven ByrnesAug 21, 2025, 5:52 PM
236 points
21 comments6 min readLW link
(x.com)

I am wor­ried about near-term non-LLM AI developments

testingthewatersJul 31, 2025, 1:15 PM
234 points
53 comments5 min readLW link

Distil­la­tion Ro­bus­tifies Unlearning

Jun 13, 2025, 1:45 PM
232 points
43 comments8 min readLW link
(arxiv.org)

Do Not Tile the Light­cone with Your Con­fused Ontology

Jan_KulveitJun 13, 2025, 12:45 PM
226 points
27 comments5 min readLW link
(boundedlyrational.substack.com)