All 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 202220232024

All Jan FebMarApr May Jun Jul Aug Sep Oct Nov Dec

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

The Waluigi Effect (mega-post)

Cleo Nardo3 Mar 2023 3:22 UTC

630 points

187 comments16 min readLW link

My Objections to “We’re All Gonna Die with Eliezer Yudkowsky”

Quintin Pope21 Mar 2023 0:06 UTC

357 points

230 comments39 min readLW link

Shutting Down the Lightcone Offices

habryka and Ben Pace

14 Mar 2023 22:47 UTC

338 points

95 comments17 min readLW link

Understanding and controlling a maze-solving policy network

TurnTrout, peligrietzer, Ulisse Mini, Monte M and David Udell

11 Mar 2023 18:59 UTC

328 points

27 comments23 min readLW link

The Parable of the King and the Random Process

moridinamael1 Mar 2023 22:18 UTC

307 points

25 comments6 min readLW link 2 reviews

Pausing AI Developments Isn’t Enough. We Need to Shut it All Down by Eliezer Yudkowsky

jacquesthibs29 Mar 2023 23:16 UTC

298 points

297 comments3 min readLW link

(time.com)

“Carefully Bootstrapped Alignment” is organizationally hard

Raemon17 Mar 2023 18:00 UTC

261 points

22 comments11 min readLW link

Discussion with Nate Soares on a key alignment difficulty

HoldenKarnofsky13 Mar 2023 21:20 UTC

256 points

42 comments22 min readLW link

Deep Deceptiveness

So8res21 Mar 2023 2:51 UTC

238 points

59 comments14 min readLW link

Natural Abstractions: Key claims, Theorems, and Critiques

LawrenceC, Leon Lang and Erik Jenner

16 Mar 2023 16:37 UTC

236 points

21 comments45 min readLW link 1 review

More information about the dangerous capability evaluations we did with GPT-4 and Claude.

Beth Barnes19 Mar 2023 0:25 UTC

233 points

54 comments8 min readLW link

(evals.alignment.org)

An AI risk argument that resonates with NYTimes readers

Julian Bradshaw12 Mar 2023 23:09 UTC

211 points

14 comments1 min readLW link

Actually, Othello-GPT Has A Linear Emergent World Representation

Neel Nanda29 Mar 2023 22:13 UTC

211 points

26 comments19 min readLW link

(neelnanda.io)

GPT-4 Plugs In

Zvi27 Mar 2023 12:10 UTC

198 points

47 comments6 min readLW link

(thezvi.wordpress.com)

Acausal normalcy

Andrew_Critch3 Mar 2023 23:34 UTC

195 points

36 comments8 min readLW link 1 review

ChatGPT (and now GPT4) is very easily distracted from its rules

dmcs15 Mar 2023 17:55 UTC

180 points

42 comments1 min readLW link

A rough and incomplete review of some of John Wentworth’s research

So8res28 Mar 2023 18:52 UTC

175 points

18 comments18 min readLW link

Why Not Just… Build Weak AI Tools For AI Alignment Research?

johnswentworth5 Mar 2023 0:12 UTC

172 points

18 comments6 min readLW link

Anthropic’s Core Views on AI Safety

Zac Hatfield-Dodds9 Mar 2023 16:55 UTC

172 points

39 comments2 min readLW link

(www.anthropic.com)

The salt in pasta water fallacy

Thomas Sepulchre27 Mar 2023 14:53 UTC

170 points

41 comments3 min readLW link

What Discovering Latent Knowledge Did and Did Not Find

Fabien Roger13 Mar 2023 19:29 UTC

166 points

17 comments11 min readLW link

Towards understanding-based safety evaluations

evhub15 Mar 2023 18:18 UTC

164 points

16 comments5 min readLW link

A stylized dialogue on John Wentworth’s claims about markets and optimization

So8res25 Mar 2023 22:32 UTC

160 points

22 comments8 min readLW link

What would a compute monitoring plan look like? [Linkpost]

Akash26 Mar 2023 19:33 UTC

158 points

10 comments4 min readLW link

(arxiv.org)

Inside the mind of a superhuman Go model: How does Leela Zero read ladders?

Haoxing Du1 Mar 2023 1:47 UTC

157 points

8 comments30 min readLW link

AI: Practical Advice for the Worried

Zvi1 Mar 2023 12:30 UTC

152 points

44 comments16 min readLW link

(thezvi.wordpress.com)

GPT-4

nz14 Mar 2023 17:02 UTC

151 points

149 comments1 min readLW link

(openai.com)

POC || GTFO culture as partial antidote to alignment wordcelism

lc15 Mar 2023 10:21 UTC

148 points

11 comments7 min readLW link

Remarks 1–18 on GPT (compressed)

Cleo Nardo20 Mar 2023 22:27 UTC

148 points

35 comments31 min readLW link

Dan Luu on “You can only communicate one top priority”

Raemon18 Mar 2023 18:55 UTC

148 points

18 comments3 min readLW link

(twitter.com)

Comments on OpenAI’s “Planning for AGI and beyond”

So8res3 Mar 2023 23:01 UTC

148 points

2 comments14 min readLW link

The Translucent Thoughts Hypotheses and Their Implications

Fabien Roger9 Mar 2023 16:30 UTC

142 points

7 comments19 min readLW link

Against LLM Reductionism

Erich_Grunewald8 Mar 2023 15:52 UTC

140 points

17 comments18 min readLW link

(www.erichgrunewald.com)

Speed running everyone through the bad alignment bingo. $5k bounty for a LW conversational agent

ArthurB9 Mar 2023 9:26 UTC

140 points

33 comments2 min readLW link

Why I’m not into the Free Energy Principle

Steven Byrnes2 Mar 2023 19:27 UTC

140 points

48 comments9 min readLW link

Why Not Just Outsource Alignment Research To An AI?

johnswentworth9 Mar 2023 21:49 UTC

139 points

49 comments9 min readLW link

Conceding a short timelines bet early

Matthew Barnett16 Mar 2023 21:49 UTC

133 points

17 comments1 min readLW link

Good News, Everyone!

jbash25 Mar 2023 13:48 UTC

131 points

23 comments2 min readLW link

We have to Upgrade

Jed McCaleb23 Mar 2023 17:53 UTC

129 points

35 comments2 min readLW link

[Linkpost] Some high-level thoughts on the DeepMind alignment team’s strategy

Vika and Rohin Shah

7 Mar 2023 11:55 UTC

128 points

13 comments5 min readLW link

(drive.google.com)

High Status Eschews Quantification of Performance

niplav19 Mar 2023 22:14 UTC

126 points

36 comments5 min readLW link

FLI open letter: Pause giant AI experiments

Zach Stein-Perlman29 Mar 2023 4:04 UTC

126 points

123 comments2 min readLW link

(futureoflife.org)

How bad a future do ML researchers expect?

KatjaGrace9 Mar 2023 4:50 UTC

122 points

8 comments2 min readLW link

(aiimpacts.org)

Manifold: If okay AGI, why?

Eliezer Yudkowsky25 Mar 2023 22:43 UTC

117 points

37 comments1 min readLW link

(manifold.markets)

ARC tests to see if GPT-4 can escape human control; GPT-4 failed to do so

Christopher King15 Mar 2023 0:29 UTC

116 points

22 comments2 min readLW link

Parasitic Language Games: maintaining ambiguity to hide conflict while burning the commons

Hazard12 Mar 2023 5:25 UTC

115 points

17 comments13 min readLW link

GPT can write Quines now (GPT-4)

Andrew_Critch14 Mar 2023 19:18 UTC

112 points

30 comments1 min readLW link

Here, have a calmness video

Kaj_Sotala16 Mar 2023 10:00 UTC

111 points

15 comments2 min readLW link

(www.youtube.com)

“Liquidity” vs “solvency” in bank runs (and some notes on Silicon Valley Bank)

rossry12 Mar 2023 9:16 UTC

107 points

27 comments12 min readLW link

The Overton Window widens: Examples of AI risk in the media

Akash23 Mar 2023 17:10 UTC

107 points

24 comments6 min readLW link