All 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 202120222023 2024 2025

All Jan Feb Mar Apr May Jun Jul Aug Sep Oct NovDec

All1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Let’s think about slowing down AI

KatjaGrace22 Dec 2022 17:40 UTC

551 points

182 comments38 min readLW link 3 reviews

(aiimpacts.org)

Staring into the abyss as a core life skill

benkuhn22 Dec 2022 15:30 UTC

347 points

22 comments12 min readLW link 1 review

(www.benkuhn.net)

Models Don’t “Get Reward”

Sam Ringer30 Dec 2022 10:37 UTC

313 points

61 comments5 min readLW link 1 review

A challenge for AGI organizations, and a challenge for readers

Rob Bensinger and Eliezer Yudkowsky

1 Dec 2022 23:11 UTC

301 points

33 comments2 min readLW link

Sazen

Duncan Sabien (Deactivated)21 Dec 2022 7:54 UTC

278 points

83 comments12 min readLW link 2 reviews

AI alignment is distinct from its near-term applications

paulfchristiano13 Dec 2022 7:10 UTC

255 points

21 comments2 min readLW link

(ai-alignment.com)

How “Discovering Latent Knowledge in Language Models Without Supervision” Fits Into a Broader Alignment Scheme

Collin15 Dec 2022 18:22 UTC

244 points

39 comments16 min readLW link 1 review

Jailbreaking ChatGPT on Release Day

Zvi2 Dec 2022 13:10 UTC

242 points

77 comments6 min readLW link 1 review

(thezvi.wordpress.com)

The Plan − 2022 Update

johnswentworth1 Dec 2022 20:43 UTC

239 points

37 comments8 min readLW link 1 review

Causal Scrubbing: a method for rigorously testing interpretability hypotheses [Redwood Research]

LawrenceC, Adrià Garriga-alonso, Nicholas Goldowsky-Dill, ryan_greenblatt, jenny, Ansh Radhakrishnan, Buck and Nate Thomas

3 Dec 2022 0:58 UTC

205 points

35 comments20 min readLW link 1 review

What AI Safety Materials Do ML Researchers Find Compelling?

Vael Gates and Collin

28 Dec 2022 2:03 UTC

175 points

34 comments2 min readLW link

The next decades might be wild

Marius Hobbhahn15 Dec 2022 16:10 UTC

175 points

42 comments41 min readLW link 1 review

Finite Factored Sets in Pictures

Magdalena Wache11 Dec 2022 18:49 UTC

174 points

35 comments12 min readLW link

Using GPT-Eliezer against ChatGPT Jailbreaking

Stuart_Armstrong and rgorman

6 Dec 2022 19:54 UTC

170 points

85 comments9 min readLW link

Things that can kill you quickly: What everyone should know about first aid

jasoncrawford27 Dec 2022 16:23 UTC

166 points

21 comments2 min readLW link 1 review

(jasoncrawford.org)

Logical induction for software engineers

Alex Flint3 Dec 2022 19:55 UTC

160 points

8 comments27 min readLW link 1 review

[Interim research report] Taking features out of superposition with sparse autoencoders

Lee Sharkey, Dan Braun and beren

13 Dec 2022 15:41 UTC

149 points

23 comments22 min readLW link 2 reviews

A Year of AI Increasing AI Progress

TW12330 Dec 2022 2:09 UTC

148 points

3 comments2 min readLW link

K-complexity is silly; use cross-entropy instead

So8res20 Dec 2022 23:06 UTC

147 points

54 comments4 min readLW link 2 reviews

Inner and outer alignment decompose one hard problem into two extremely hard problems

TurnTrout2 Dec 2022 2:43 UTC

147 points

22 comments47 min readLW link 3 reviews

Updating my AI timelines

Matthew Barnett5 Dec 2022 20:46 UTC

145 points

50 comments2 min readLW link

Shard Theory in Nine Theses: a Distillation and Critical Appraisal

LawrenceC19 Dec 2022 22:52 UTC

143 points

30 comments18 min readLW link

[Question] How to Convince my Son that Drugs are Bad

concerned_dad17 Dec 2022 18:47 UTC

140 points

84 comments2 min readLW link

Re-Examining LayerNorm

Eric Winsor1 Dec 2022 22:20 UTC

127 points

12 comments5 min readLW link

Shared reality: a key driver of human behavior

kdbscott24 Dec 2022 19:35 UTC

126 points

25 comments4 min readLW link

Deconfusing Direct vs Amortised Optimization

beren2 Dec 2022 11:30 UTC

124 points

19 comments10 min readLW link

Did ChatGPT just gaslight me?

TW1231 Dec 2022 5:41 UTC

123 points

45 comments9 min readLW link

(aiwatchtower.substack.com)

The case against AI alignment

andrew sauer24 Dec 2022 6:57 UTC

118 points

110 comments5 min readLW link

[Question] Why The Focus on Expected Utility Maximisers?

DragonGod27 Dec 2022 15:49 UTC

118 points

84 comments3 min readLW link

200 Concrete Open Problems in Mechanistic Interpretability: Introduction

Neel Nanda28 Dec 2022 21:06 UTC

106 points

0 comments10 min readLW link

Trying to disambiguate different questions about whether RLHF is “good”

Buck14 Dec 2022 4:03 UTC

106 points

47 comments7 min readLW link 1 review

Language models are nearly AGIs but we don’t notice it because we keep shifting the bar

philosophybear30 Dec 2022 5:15 UTC

105 points

13 comments7 min readLW link

Slightly against aligning with neo-luddites

Matthew Barnett26 Dec 2022 22:46 UTC

104 points

31 comments4 min readLW link

But is it really in Rome? An investigation of the ROME model editing technique

jacquesthibs30 Dec 2022 2:40 UTC

104 points

2 comments18 min readLW link

[Linkpost] The Story Of VaccinateCA

hath9 Dec 2022 23:54 UTC

103 points

4 comments10 min readLW link

(www.worksinprogress.co)

Applied Linear Algebra Lecture Series

johnswentworth22 Dec 2022 6:57 UTC

103 points

8 comments1 min readLW link

Thoughts on AGI organizations and capabilities work

Rob Bensinger and So8res

7 Dec 2022 19:46 UTC

102 points

17 comments5 min readLW link

Finding gliders in the game of life

paulfchristiano1 Dec 2022 20:40 UTC

101 points

8 comments16 min readLW link

(ai-alignment.com)

Discovering Language Model Behaviors with Model-Written Evaluations

evhub and Ethan Perez

20 Dec 2022 20:08 UTC

100 points

34 comments1 min readLW link

(www.anthropic.com)

Bad at Arithmetic, Promising at Math

cohenmacaulay18 Dec 2022 5:40 UTC

100 points

19 comments20 min readLW link 1 review

[Link] Why I’m optimistic about OpenAI’s alignment approach

janleike5 Dec 2022 22:51 UTC

98 points

15 comments1 min readLW link

(aligned.substack.com)

You can still fetch the coffee today if you’re dead tomorrow

davidad9 Dec 2022 14:06 UTC

96 points

19 comments5 min readLW link

The LessWrong 2021 Review: Intellectual Circle Expansion

Ruby and Raemon

1 Dec 2022 21:17 UTC

95 points

55 comments8 min readLW link

Revisiting algorithmic progress

Tamay and Ege Erdil

13 Dec 2022 1:39 UTC

95 points

15 comments2 min readLW link 1 review

(arxiv.org)

Towards Hodge-podge Alignment

Cleo Nardo19 Dec 2022 20:12 UTC

95 points

30 comments9 min readLW link

A Comprehensive Mechanistic Interpretability Explainer & Glossary

Neel Nanda21 Dec 2022 12:35 UTC

91 points

6 comments2 min readLW link

(neelnanda.io)

Setting the Zero Point

Duncan Sabien (Deactivated)9 Dec 2022 6:06 UTC

91 points

43 comments20 min readLW link 1 review

Local Memes Against Geometric Rationality

Scott Garrabrant21 Dec 2022 3:53 UTC

90 points

3 comments6 min readLW link

Can we efficiently distinguish different mechanisms?

paulfchristiano27 Dec 2022 0:20 UTC

88 points

30 comments16 min readLW link

(ai-alignment.com)

Consider using reversible automata for alignment research

Alex_Altair11 Dec 2022 1:00 UTC

88 points

30 comments2 min readLW link