14 Mar 2023 22:47 UTC

337 points

93 comments17 min readLW link

[Question] What are some ideas that LessWrong has reinvented?

RomanHauksson14 Mar 2023 22:27 UTC

4 points

13 comments1 min readLW link

Human preferences as RL critic values—implications for alignment

Seth Herd14 Mar 2023 22:10 UTC

21 points

6 comments6 min readLW link

PaperclipGPT(-4)

Michael Tontchev14 Mar 2023 22:03 UTC

7 points

0 comments11 min readLW link

GPT-4 developer livestream

Gerald Monroe14 Mar 2023 20:55 UTC

9 points

0 comments1 min readLW link

(www.youtube.com)

[Question] Main actors in the AI race

Marta14 Mar 2023 20:50 UTC

3 points

1 comment1 min readLW link

Success without dignity: a nearcasting story of avoiding catastrophe by luck

HoldenKarnofsky14 Mar 2023 19:23 UTC

74 points

8 comments15 min readLW link

GPT can write Quines now (GPT-4)

Andrew_Critch14 Mar 2023 19:18 UTC

111 points

30 comments1 min readLW link

Vector semantics and the (in-context) construction of meaning in Coleridge’s “Kubla Khan”

Bill Benzon14 Mar 2023 19:16 UTC

4 points

0 comments7 min readLW link

A better analogy and example for teaching AI takeover: the ML Inferno

Christopher King14 Mar 2023 19:14 UTC

18 points

0 comments5 min readLW link

PaLM API & MakerSuite

Gabriel Mukobi14 Mar 2023 19:08 UTC

20 points

1 comment1 min readLW link

(developers.googleblog.com)

What is a definition, how can it be extrapolated?

Stuart_Armstrong14 Mar 2023 18:08 UTC

34 points

5 comments7 min readLW link

Cambridge LW: Rationality Practice: The Map is Not the Territory

Darmani14 Mar 2023 17:56 UTC

6 points

0 comments1 min readLW link

[Question] Beneficial initial conditions for AGI

mikbp14 Mar 2023 17:41 UTC

1 point

3 comments1 min readLW link

[Question] “The elephant in the room: the biggest risk of artificial intelligence may not be what we think” What to say about that?

Obladi Oblada14 Mar 2023 17:37 UTC

−5 points

0 comments3 min readLW link

GPT-4

nz14 Mar 2023 17:02 UTC

150 points

149 comments1 min readLW link

(openai.com)

Storytelling Makes GPT-3.5 Deontologist: Unexpected Effects of Context on LLM Behavior

Edmund Mills and Scott Emmons

14 Mar 2023 8:44 UTC

17 points

0 comments12 min readLW link

Forecasting Authoritarian and Sovereign Power uses of Large Language Models

K. Liam Smith14 Mar 2023 8:44 UTC

7 points

0 comments8 min readLW link

(taboo.substack.com)

Fixed points in mortal population games

ViktoriaMalyasova14 Mar 2023 7:10 UTC

24 points

0 comments12 min readLW link

(www.lesswrong.com)

To determine alignment difficulty, we need to know the absolute difficulty of alignment generalization

Jeffrey Ladish14 Mar 2023 3:52 UTC

12 points

3 comments2 min readLW link

EA & LW Forum Weekly Summary (6th − 12th March 2023)

Zoe Williams14 Mar 2023 3:01 UTC

7 points

0 comments1 min readLW link

Alpaca: A Strong Open-Source Instruction-Following Model

sanxiyn14 Mar 2023 2:41 UTC

26 points

2 comments1 min readLW link

(crfm.stanford.edu)

Discussion with Nate Soares on a key alignment difficulty

HoldenKarnofsky13 Mar 2023 21:20 UTC

250 points

38 comments22 min readLW link

What Discovering Latent Knowledge Did and Did Not Find

Fabien Roger13 Mar 2023 19:29 UTC

164 points

16 comments11 min readLW link

South Bay ACX/LW Meetup

IS13 Mar 2023 18:25 UTC

2 points

0 comments1 min readLW link

Could Roko’s basilisk acausally bargain with a paperclip maximizer?

Christopher King13 Mar 2023 18:21 UTC

1 point

8 comments1 min readLW link

Bayesian optimization to find molecules that bind to proteins

rotatingpaguro13 Mar 2023 18:17 UTC

1 point

0 comments1 min readLW link

(www.youtube.com)

Linkpost: ‘Dissolving’ AI Risk – Parameter Uncertainty in AI Future Forecasting

DavidW13 Mar 2023 16:52 UTC

6 points

0 comments1 min readLW link

(forum.effectivealtruism.org)

Decentralized Exclusion

jefftk13 Mar 2023 15:50 UTC

23 points

19 comments2 min readLW link

(www.jefftk.com)

Linkpost: A Contra AI FOOM Reading List

DavidW13 Mar 2023 14:45 UTC

25 points

4 comments1 min readLW link

(magnusvinding.com)

Linkpost: A tale of 2.5 orthogonality theses

DavidW13 Mar 2023 14:19 UTC

9 points

3 comments1 min readLW link

(forum.effectivealtruism.org)

Plan for mediocre alignment of brain-like [model-based RL] AGI

Steven Byrnes13 Mar 2023 14:11 UTC

63 points

24 comments12 min readLW link

your terminal values are complex and not objective

Tamsin Leake13 Mar 2023 13:34 UTC

60 points

6 comments2 min readLW link

(carado.moe)

Against AGI Timelines

Jonathan Yan13 Mar 2023 13:33 UTC

13 points

3 comments1 min readLW link

(benlandautaylor.com)

What is calibration?

AlexMennen13 Mar 2023 6:30 UTC

27 points

1 comment4 min readLW link

On taking AI risk seriously

Eleni Angelou13 Mar 2023 5:50 UTC

6 points

0 comments1 min readLW link

(www.nytimes.com)

Nose / throat treatments for respiratory infections

juliawise13 Mar 2023 2:41 UTC

46 points

6 comments8 min readLW link

Gold, Silver, Red: A color scheme for understanding people

Michael Soareverix13 Mar 2023 1:06 UTC

17 points

2 comments4 min readLW link

Yudkowsky on AGI risk on the Bankless podcast

Rob Bensinger13 Mar 2023 0:42 UTC

83 points

5 comments1 min readLW link

Thoughts on self-inspecting neural networks.

Deruwyn12 Mar 2023 23:58 UTC

4 points

2 comments5 min readLW link

An AI risk argument that resonates with NYTimes readers

Julian Bradshaw12 Mar 2023 23:09 UTC

203 points

14 comments1 min readLW link

Musicians and Mouths

jefftk12 Mar 2023 22:50 UTC

13 points

7 comments2 min readLW link

(www.jefftk.com)

Are there cognitive realms?

TsviBT12 Mar 2023 19:28 UTC

25 points

2 comments10 min readLW link

[Question] What happened on the Extropians message board?

politicalpersuasion12 Mar 2023 19:22 UTC

−53 points

1 comment1 min readLW link

Creating a Discord server for Mechanistic Interpretability Projects

Victor Levoso12 Mar 2023 18:00 UTC

30 points

6 comments2 min readLW link

Paper Replication Walkthrough: Reverse-Engineering Modular Addition

Neel Nanda12 Mar 2023 13:25 UTC

18 points

0 comments1 min readLW link

(neelnanda.io)

the quantum amplitude argument against ethics deduplication

Tamsin Leake12 Mar 2023 13:02 UTC

11 points

16 comments2 min readLW link

(carado.moe)

What problems do African-Americans face? An initial investigation using Standpoint Epistemology and Surveys

tailcalled12 Mar 2023 11:42 UTC

41 points

26 comments15 min readLW link

“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

“You’ll Never Persuade People Like That”

Zack_M_Davis12 Mar 2023 5:38 UTC

14 points

31 comments2 min readLW link