12 Dec 2023 18:14 UTC

436 points

189 comments33 min readLW link

Speaking to Congressional staffers about AI risk

Akash and hath

4 Dec 2023 23:08 UTC

298 points

23 comments15 min readLW link

Constellations are Younger than Continents

Jeffrey Heninger19 Dec 2023 6:12 UTC

260 points

22 comments2 min readLW link

AI Control: Improving Safety Despite Intentional Subversion

Buck, Fabien Roger, ryan_greenblatt and Kshitij Sachan

13 Dec 2023 15:51 UTC

228 points

18 comments10 min readLW link 1 review

Thoughts on “AI is easy to control” by Pope & Belrose

Steven Byrnes1 Dec 2023 17:30 UTC

197 points

64 comments14 min readLW link 1 review

“Humanity vs. AGI” Will Never Look Like “Humanity vs. AGI” to Humanity

Thane Ruthenis16 Dec 2023 20:08 UTC

189 points

34 comments5 min readLW link

re: Yudkowsky on biological materials

bhauth11 Dec 2023 13:28 UTC

180 points

30 comments5 min readLW link

Effective Aspersions: How the Nonlinear Investigation Went Wrong

TracingWoodgrains19 Dec 2023 12:00 UTC

179 points

171 comments1 min readLW link 1 review

Critical review of Christiano’s disagreements with Yudkowsky

Vanessa Kosoy27 Dec 2023 16:02 UTC

172 points

40 comments15 min readLW link

Is being sexy for your homies?

Valentine13 Dec 2023 20:37 UTC

169 points

97 comments14 min readLW link 2 reviews

2023 Unofficial LessWrong Census/Survey

Screwtape2 Dec 2023 4:41 UTC

169 points

81 comments1 min readLW link

The ‘Neglected Approaches’ Approach: AE Studio’s Alignment Agenda

Cameron Berg, Judd Rosenblatt, AE Studio and Marc Carauleanu

18 Dec 2023 20:35 UTC

168 points

21 comments12 min readLW link

How useful is mechanistic interpretability?

ryan_greenblatt, Neel Nanda, Buck and habryka

1 Dec 2023 2:54 UTC

165 points

54 comments25 min readLW link

The likely first longevity drug is based on sketchy science. This is bad for science and bad for longevity.

BobBurgers12 Dec 2023 2:42 UTC

161 points

34 comments5 min readLW link

Succession

Richard_Ngo20 Dec 2023 19:25 UTC

159 points

48 comments11 min readLW link

(www.narrativeark.xyz)

Most People Don’t Realize We Have No Idea How Our AIs Work

Thane Ruthenis21 Dec 2023 20:02 UTC

158 points

42 comments1 min readLW link

Discussion: Challenges with Unsupervised LLM Knowledge Discovery

Seb Farquhar, Vikrant Varma, zac_kenton, gasteigerjo, Vlad Mikulik and Rohin Shah

18 Dec 2023 11:58 UTC

147 points

21 comments10 min readLW link

The Plan − 2023 Version

johnswentworth29 Dec 2023 23:34 UTC

146 points

40 comments31 min readLW link

AI Views Snapshots

Rob Bensinger13 Dec 2023 0:45 UTC

142 points

61 comments1 min readLW link

The Dark Arts

lsusr and Lyrongolem

19 Dec 2023 4:41 UTC

132 points

49 comments9 min readLW link

Current AIs Provide Nearly No Data Relevant to AGI Alignment

Thane Ruthenis15 Dec 2023 20:16 UTC

124 points

156 comments8 min readLW link

Bayesian Injustice

Kevin Dorst14 Dec 2023 15:44 UTC

124 points

10 comments6 min readLW link

(kevindorst.substack.com)

Deep Forgetting & Unlearning for Safely-Scoped LLMs

scasper5 Dec 2023 16:48 UTC

123 points

30 comments13 min readLW link

Natural Latents: The Math

johnswentworth and David Lorell

27 Dec 2023 19:03 UTC

120 points

37 comments12 min readLW link

The LessWrong 2022 Review

habryka5 Dec 2023 4:00 UTC

115 points

43 comments4 min readLW link

Mapping the semantic void: Strange goings-on in GPT embedding spaces

mwatkins14 Dec 2023 13:10 UTC

114 points

31 comments14 min readLW link

What I Would Do If I Were Working On AI Governance

johnswentworth8 Dec 2023 6:43 UTC

110 points

32 comments10 min readLW link

Fact Finding: Attempting to Reverse-Engineer Factual Recall on the Neuron Level (Post 1)

Neel Nanda, Senthooran Rajamanoharan, János Kramár and Rohin Shah

23 Dec 2023 2:44 UTC

108 points

9 comments22 min readLW link 1 review

“AI Alignment” is a Dangerously Overloaded Term

Roko15 Dec 2023 14:34 UTC

108 points

100 comments3 min readLW link

[Question] How do you feel about LessWrong these days? [Open feedback thread]

jacobjacob5 Dec 2023 20:54 UTC

107 points

284 comments1 min readLW link

On the future of language models

owencb20 Dec 2023 16:58 UTC

105 points

17 comments1 min readLW link

Nonlinear’s Evidence: Debunking False and Misleading Claims

KatWoods12 Dec 2023 13:16 UTC

104 points

171 comments1 min readLW link

The Witness

Richard_Ngo3 Dec 2023 22:27 UTC

104 points

5 comments14 min readLW link

(www.narrativeark.xyz)

[Valence series] 1. Introduction

Steven Byrnes4 Dec 2023 15:40 UTC

98 points

16 comments16 min readLW link 2 reviews

Meaning & Agency

abramdemski19 Dec 2023 22:27 UTC

91 points

17 comments14 min readLW link

Prediction Markets aren’t Magic

SimonM21 Dec 2023 12:54 UTC

90 points

29 comments3 min readLW link

Nietzsche’s Morality in Plain English

Arjun Panickssery4 Dec 2023 0:57 UTC

90 points

14 comments4 min readLW link 1 review

(arjunpanickssery.substack.com)

Based Beff Jezos and the Accelerationists

Zvi6 Dec 2023 16:00 UTC

89 points

29 comments12 min readLW link

(thezvi.wordpress.com)

A Crisper Explanation of Simulacrum Levels

Thane Ruthenis23 Dec 2023 22:13 UTC

89 points

13 comments13 min readLW link

[Valence series] 2. Valence & Normativity

Steven Byrnes7 Dec 2023 16:43 UTC

88 points

6 comments28 min readLW link

Some for-profit AI alignment org ideas

Eric Ho14 Dec 2023 14:23 UTC

86 points

19 comments9 min readLW link

A Universal Emergent Decomposition of Retrieval Tasks in Language Models

Alexandre Variengien and Eric Winsor

19 Dec 2023 11:52 UTC

84 points

3 comments10 min readLW link

(arxiv.org)

Refusal mechanisms: initial experiments with Llama-2-7b-chat

Andy Arditi and Oscar Obeso

8 Dec 2023 17:08 UTC

81 points

7 comments7 min readLW link

Studying The Alien Mind

Quentin FEUILLADE--MONTIXI and NicholasKees

5 Dec 2023 17:27 UTC

80 points

10 comments15 min readLW link

EU policymakers reach an agreement on the AI Act

tlevin15 Dec 2023 6:02 UTC

78 points

7 comments7 min readLW link

Send us example gnarly bugs

Beth Barnes, Megan Kinniment and Tao Lin

10 Dec 2023 5:23 UTC

77 points

10 comments2 min readLW link

MATS Summer 2023 Retrospective

utilistrutil, Juan Gil, Ryan Kidd, Christian Smith, McKennaFitzgerald and LauraVaughan

1 Dec 2023 23:29 UTC

77 points

34 comments26 min readLW link

OpenAI: Leaks Confirm the Story

Zvi12 Dec 2023 14:00 UTC

77 points

9 comments16 min readLW link

(thezvi.wordpress.com)

[Valence series] 3. Valence & Beliefs

Steven Byrnes11 Dec 2023 20:21 UTC

75 points

11 comments21 min readLW link

The problems with the concept of an infohazard as used by the LW community [Linkpost]

Noosphere8922 Dec 2023 16:13 UTC

75 points

43 comments3 min readLW link

(www.beren.io)