At­tach Your­self to the Right Per­son, and You’ll Go Far (a nerdy poem about bugs)

Character#273615 Feb 2026 23:36 UTC
10 points
0 comments1 min readLW link

Model mul­ti­task­ing: Can a model learn two differ­ent tasks si­mul­ta­neously through Grokking?

arcee18315 Feb 2026 23:06 UTC
7 points
0 comments9 min readLW link

Phan­tom Trans­fer and the Ba­sic Science of Data Poisoning

15 Feb 2026 19:51 UTC
82 points
8 comments6 min readLW link

Should any­one’s “anal­y­sis” of ex­tremely com­plex sys­tems, such as geopoli­tics, be taken se­ri­ously? or, Does any­one take a 5 year old’s “anal­y­sis” of de­cently com­plex sys­tems, like big city poli­tics, se­ri­ously?

M. Y. Zuo15 Feb 2026 18:44 UTC
20 points
5 comments1 min readLW link

Pain­less Ac­ti­va­tion Steering

Sasha Cui15 Feb 2026 17:49 UTC
14 points
2 comments1 min readLW link
(open.substack.com)

PieArena: Lan­guage Agents Ne­go­ti­at­ing Against Yale MBAs

Sasha Cui15 Feb 2026 17:45 UTC
5 points
0 comments1 min readLW link
(open.substack.com)

The Miss­ing Se­quence: Why Cor­rect Anal­y­sis Makes Ter­rible Ac­tion Guides

Florian_Dietz15 Feb 2026 17:24 UTC
15 points
6 comments6 min readLW link

The Friendly Telepath Problems

Gunnar_Zarncke15 Feb 2026 15:08 UTC
31 points
9 comments7 min readLW link

Were witches in­fer­tile men­tally ill women?

Yair Halberstadt15 Feb 2026 13:20 UTC
20 points
5 comments1 min readLW link

Con­tra Alexan­der’s Half-Defence of Bio Anchors

J Bostock15 Feb 2026 13:08 UTC
38 points
3 comments7 min readLW link

Real­iz­abil­ity for Finite State Re­ac­tive Agents

Vardhan15 Feb 2026 11:38 UTC
10 points
4 comments28 min readLW link

Words That Belong to Someone

Wen Xing15 Feb 2026 5:53 UTC
17 points
3 comments3 min readLW link

It’s Not That Hard to Build Re­cur­sively Self-Im­prov­ing AI

Charlie Sanders15 Feb 2026 4:35 UTC
11 points
8 comments2 min readLW link

AXRP Epi­sode 48 - Guive As­sadi on AI Prop­erty Rights

DanielFilan15 Feb 2026 2:20 UTC
22 points
0 comments75 min readLW link

Another short cri­tique of the An­thropic “Hot Mess” paper

i_am_nuts15 Feb 2026 1:38 UTC
9 points
0 comments2 min readLW link
(iamnuts.substack.com)

Inoc­u­la­tion Prompt­ing: Open Ques­tions and My Re­search Priorities

henryc15 Feb 2026 0:16 UTC
12 points
0 comments4 min readLW link

sunsbeams

Jared M.14 Feb 2026 21:19 UTC
4 points
0 comments2 min readLW link

Im­mor­tal­ity: A Begin­ner’s Guide (Part 1)

MarkelKori14 Feb 2026 21:17 UTC
17 points
4 comments3 min readLW link

The Wor­thy Inheritor

Bridgett Kay14 Feb 2026 19:10 UTC
−1 points
0 comments8 min readLW link
(dxmrevealed.wordpress.com)

A multi-level post­mortem of how our whole house got badly poisoned

Lucie Philippon14 Feb 2026 16:11 UTC
76 points
5 comments7 min readLW link
(aelerinya.substack.com)

LLMs strug­gle to ver­bal­ize their in­ter­nal reasoning

Emil Ryd14 Feb 2026 15:32 UTC
50 points
9 comments9 min readLW link

De­liber­ate Epistemic Uncer­tainty: An Au­to­mated Ex­per­i­ment on AI Self-Reporting

Florian_Dietz14 Feb 2026 15:13 UTC
13 points
0 comments8 min readLW link

LessWrong Is Sleep­ing On In­ter­net Cul­ture Anal­y­sis – And So Is The Rest Of The Web

Bowl of Cereal14 Feb 2026 14:58 UTC
−3 points
0 comments1 min readLW link

Beloved by Chatbots

Ben14 Feb 2026 12:12 UTC
22 points
7 comments3 min readLW link

Life at the Frontlines of De­mo­graphic Collapse

Martin Sustrik14 Feb 2026 6:30 UTC
289 points
52 comments8 min readLW link
(www.250bpm.com)

Ads, In­cen­tives, and Destiny

Against Moloch14 Feb 2026 5:41 UTC
31 points
1 comment4 min readLW link
(againstmoloch.com)

Why I’m Wor­ried About Job Loss + Thoughts on Com­par­a­tive Advantage

claywren13 Feb 2026 23:36 UTC
62 points
5 comments11 min readLW link

METR Time Hori­zons: Now 10x/​Year

johncrox13 Feb 2026 23:01 UTC
28 points
6 comments3 min readLW link

Use more text than one to­ken to avoid neuralese

Jude Stiel13 Feb 2026 21:09 UTC
10 points
4 comments1 min readLW link

Hazards of Selec­tion Effects on Ap­proved Information

Zack_M_Davis13 Feb 2026 18:51 UTC
56 points
11 comments12 min readLW link
(zackmdavis.net)

OpenClaw Newsletter

Jacobson13 Feb 2026 17:59 UTC
2 points
1 comment5 min readLW link

ChatGPT-5.3-Codex Is Also Good At Coding

Zvi13 Feb 2026 16:20 UTC
45 points
2 comments20 min readLW link
(thezvi.wordpress.com)

Repli­ca­tion of Koorndijk (2025): Differ­en­tial Com­pli­ance May Reflect Prompt Sen­si­tivity Rather Than Strate­gic Reasoning

13 Feb 2026 16:12 UTC
9 points
0 comments8 min readLW link

Towards an ob­jec­tive test of Com­pas­sion—Turn­ing an ab­stract test into a col­lec­tion of nuances

tailcalled13 Feb 2026 15:03 UTC
12 points
0 comments7 min readLW link

(Up­dated) METR’s data can’t dis­t­in­guish be­tween tra­jec­to­ries (and 80% hori­zons are an or­der of mag­ni­tude off)

Jonas Moss13 Feb 2026 14:05 UTC
28 points
10 comments10 min readLW link

We Die Be­cause it’s a Com­pu­ta­tional Necessity

E.G. Blee-Goldman13 Feb 2026 13:16 UTC
2 points
3 comments22 min readLW link

Hazardous States and Accidents

kqr13 Feb 2026 13:02 UTC
4 points
0 comments4 min readLW link
(entropicthoughts.com)

Sys­temic Risks and Where to Find Them

Jonas Hallgren13 Feb 2026 10:51 UTC
14 points
0 comments20 min readLW link
(equilibria1.substack.com)

Nick Bostrom: Op­ti­mal Timing for Superintelligence

Julian Bradshaw13 Feb 2026 7:33 UTC
8 points
3 comments2 min readLW link
(nickbostrom.com)

Why You Don’t Believe in Xhosa Prophecies

Jan_Kulveit13 Feb 2026 2:25 UTC
265 points
28 comments4 min readLW link

Gem­ini’s Hy­po­thet­i­cal Present

jefftk13 Feb 2026 2:20 UTC
101 points
9 comments2 min readLW link
(www.jefftk.com)

I Tried to Trick My­self into Be­ing a Bet­ter Plan­ner & Prob­lem Solver

CstineSublime13 Feb 2026 0:25 UTC
7 points
2 comments3 min readLW link

Grad­ing AI 2027′s 2025 Predictions

13 Feb 2026 0:18 UTC
64 points
4 comments9 min readLW link
(blog.ai-futures.org)

Long-term risks from ide­olog­i­cal fanaticism

12 Feb 2026 23:26 UTC
99 points
12 comments84 min readLW link

(Re)Dis­cov­er­ing Nat­u­ral Laws

Margot12 Feb 2026 21:45 UTC
13 points
0 comments17 min readLW link

An On­tol­ogy of Rep­re­sen­ta­tions: Limits of Universality

Margot12 Feb 2026 21:43 UTC
23 points
1 comment39 min readLW link

A Closer Look at the “So­cieties of Thought” Paper

Against Moloch12 Feb 2026 21:38 UTC
10 points
0 comments3 min readLW link
(againstmoloch.com)

mod­els have some pretty funny at­trac­tor states

12 Feb 2026 21:14 UTC
275 points
38 comments18 min readLW link

Stay in your hu­man loop

benjamin ar12 Feb 2026 21:05 UTC
22 points
0 comments5 min readLW link
(bjar.substack.com)

The case for in­dus­trial evals

12 Feb 2026 20:45 UTC
16 points
0 comments23 min readLW link