5 Feb 2023 22:02 UTC

663 points

204 comments12 min readLW link

Focus on the places where you feel shocked everyone’s dropping the ball

So8res2 Feb 2023 0:27 UTC

413 points

61 comments4 min readLW link

Bing Chat is blatantly, aggressively misaligned

evhub15 Feb 2023 5:29 UTC

396 points

168 comments2 min readLW link

Noting an error in Inadequate Equilibria

Matthew Barnett8 Feb 2023 1:33 UTC

359 points

56 comments2 min readLW link

Please don’t throw your mind away

TsviBT15 Feb 2023 21:41 UTC

336 points

44 comments18 min readLW link

Cyborgism

NicholasKees and janus

10 Feb 2023 14:47 UTC

333 points

45 comments35 min readLW link

Childhoods of exceptional people

Henrik Karlsson6 Feb 2023 17:27 UTC

324 points

62 comments15 min readLW link

(escapingflatland.substack.com)

Fucking Goddamn Basics of Rationalist Discourse

LoganStrohl4 Feb 2023 1:47 UTC

301 points

97 comments1 min readLW link

I hired 5 people to sit behind me and make me productive for a month

Simon Berens5 Feb 2023 1:19 UTC

243 points

81 comments10 min readLW link

(www.simonberens.com)

AGI in sight: our look at the game board

Andrea_Miotti and Gabriel Alfour

18 Feb 2023 22:17 UTC

227 points

135 comments6 min readLW link

(andreamiotti.substack.com)

You Don’t Exist, Duncan

[DEACTIVATED] Duncan Sabien2 Feb 2023 8:37 UTC

227 points

107 comments9 min readLW link

Elements of Rationalist Discourse

Rob Bensinger12 Feb 2023 7:58 UTC

215 points

47 comments3 min readLW link

Cognitive Emulation: A Naive AI Safety Proposal

Connor Leahy and Gabriel Alfour

25 Feb 2023 19:35 UTC

197 points

45 comments4 min readLW link

AI alignment researchers don’t (seem to) stack

So8res21 Feb 2023 0:48 UTC

188 points

40 comments3 min readLW link

EigenKarma: trust at scale

Henrik Karlsson8 Feb 2023 18:52 UTC

182 points

50 comments5 min readLW link

Why Are Bacteria So Simple?

aysja6 Feb 2023 3:00 UTC

171 points

33 comments10 min readLW link

AI #1: Sydney and Bing

Zvi21 Feb 2023 14:00 UTC

170 points

44 comments61 min readLW link

(thezvi.wordpress.com)

Parametrically retargetable decision-makers tend to seek power

TurnTrout18 Feb 2023 18:41 UTC

166 points

9 comments2 min readLW link

(arxiv.org)

My understanding of Anthropic strategy

Swimmer963 (Miranda Dixon-Luinenburg) 15 Feb 2023 1:56 UTC

165 points

31 comments4 min readLW link

[Link] A community alert about Ziz

DanielFilan24 Feb 2023 0:06 UTC

163 points

124 comments2 min readLW link

(medium.com)

Big Mac Subsidy?

jefftk23 Feb 2023 4:00 UTC

154 points

24 comments2 min readLW link

(www.jefftk.com)

We Found An Neuron in GPT-2

Joseph Miller and Clement Neo

11 Feb 2023 18:27 UTC

141 points

22 comments7 min readLW link

(clementneo.com)

Stop posting prompt injections on Twitter and calling it “misalignment”

lc19 Feb 2023 2:21 UTC

138 points

9 comments1 min readLW link

Full Transcript: Eliezer Yudkowsky on the Bankless podcast

remember and Andrea_Miotti

23 Feb 2023 12:34 UTC

138 points

89 comments75 min readLW link

Anomalous tokens reveal the original identities of Instruct models

janus and jdp

9 Feb 2023 1:30 UTC

136 points

16 comments9 min readLW link

(generative.ink)

Modal Fixpoint Cooperation without Löb’s Theorem

Andrew_Critch5 Feb 2023 0:58 UTC

133 points

32 comments3 min readLW link

Pretraining Language Models with Human Preferences

Tomek Korbak, Sam Bowman and Ethan Perez

21 Feb 2023 17:57 UTC

133 points

18 comments11 min readLW link

“Rationalist Discourse” Is Like “Physicist Motors”

Zack_M_Davis26 Feb 2023 5:58 UTC

131 points

152 comments9 min readLW link

Evaluations (of new AI Safety researchers) can be noisy

LawrenceC5 Feb 2023 4:15 UTC

130 points

10 comments16 min readLW link

Hashing out long-standing disagreements seems low-value to me

So8res16 Feb 2023 6:20 UTC

130 points

34 comments4 min readLW link

Recommendation: Bug Bounties and Responsible Disclosure for Advanced ML Systems

Vaniver17 Feb 2023 20:11 UTC

124 points

11 comments2 min readLW link

There are (probably) no superhuman Go AIs: strong human players beat the strongest AIs

Taran19 Feb 2023 12:25 UTC

123 points

33 comments4 min readLW link

In Defense of Chatbot Romance

Kaj_Sotala11 Feb 2023 14:30 UTC

123 points

52 comments11 min readLW link

(kajsotala.fi)

A proposed method for forecasting transformative AI

Matthew Barnett10 Feb 2023 19:34 UTC

121 points

20 comments10 min readLW link

There are no coherence theorems

Dan H and EJT

20 Feb 2023 21:25 UTC

121 points

114 comments19 min readLW link

One-layer transformers aren’t equivalent to a set of skip-trigrams

Buck17 Feb 2023 17:26 UTC

119 points

10 comments7 min readLW link

GPT-175bee

Adam Scherlis and LawrenceC

8 Feb 2023 18:58 UTC

119 points

13 comments1 min readLW link

On Investigating Conspiracy Theories

Zvi20 Feb 2023 12:50 UTC

115 points

38 comments5 min readLW link

(thezvi.wordpress.com)

The public supports regulating AI for safety

Zach Stein-Perlman17 Feb 2023 4:10 UTC

114 points

9 comments1 min readLW link

(aiimpacts.org)

The Open Agency Model

Eric Drexler22 Feb 2023 10:35 UTC

113 points

18 comments4 min readLW link

Bing chat is the AI fire alarm

Ratios17 Feb 2023 6:51 UTC

112 points

62 comments3 min readLW link

GPT-4 Predictions

Stephen McAleese17 Feb 2023 23:20 UTC

109 points

27 comments11 min readLW link

SolidGoldMagikarp II: technical details and more recent findings

mwatkins and Jessica Rumbelow

6 Feb 2023 19:09 UTC

109 points

45 comments13 min readLW link

A Way To Be Okay

[DEACTIVATED] Duncan Sabien19 Feb 2023 20:27 UTC

107 points

36 comments10 min readLW link

Conflict Theory of Bounded Distrust

Zack_M_Davis12 Feb 2023 5:30 UTC

106 points

29 comments3 min readLW link

I don’t think MIRI “gave up”

Raemon3 Feb 2023 0:26 UTC

105 points

64 comments4 min readLW link

Sam Altman: “Planning for AGI and beyond”

LawrenceC24 Feb 2023 20:28 UTC

104 points

54 comments6 min readLW link

(openai.com)

Cyborg Periods: There will be multiple AI transitions

Jan_Kulveit and rosehadshar

22 Feb 2023 16:09 UTC

103 points

9 comments6 min readLW link

Don’t accelerate problems you’re trying to solve

Andrea_Miotti and remember

15 Feb 2023 18:11 UTC

100 points

26 comments4 min readLW link

H5N1

Zvi13 Feb 2023 12:50 UTC

100 points

1 comment9 min readLW link

(thezvi.wordpress.com)