[Question] What are the most com­mon so­cial in­se­cu­ri­ties?

ChipmonkJan 16, 2024, 5:24 PM
9 points
6 comments1 min readLW link

Why wasn’t preser­va­tion with the goal of po­ten­tial fu­ture re­vival started ear­lier in his­tory?

Andy_McKenzieJan 16, 2024, 4:15 PM
31 points
1 comment6 min readLW link

[Question] Why are peo­ple un­keen to im­mor­tal­ity that would come from tech­nolog­i­cal ad­vance­ments and/​or AI?

Gabi QUENEJan 16, 2024, 2:23 PM
12 points
41 comments1 min readLW link

Deal­ing with Awkwardness

Jonathan MoregårdJan 16, 2024, 12:32 PM
13 points
0 comments4 min readLW link
(honestliving.substack.com)

The im­pos­si­ble prob­lem of due process

mingyuanJan 16, 2024, 5:18 AM
197 points
64 comments14 min readLW link

[Re­tracted] New­ton’s law of cool­ing from first principles

NisanJan 16, 2024, 4:21 AM
9 points
15 comments2 min readLW link

Sparse Au­toen­coders Work on At­ten­tion Layer Outputs

Jan 16, 2024, 12:26 AM
85 points
9 comments18 min readLW link

Goals se­lected from learned knowl­edge: an al­ter­na­tive to RL alignment

Seth HerdJan 15, 2024, 9:52 PM
42 points
18 comments7 min readLW link

In­tro­duc­ing REBUS: A Ro­bust Eval­u­a­tion Bench­mark of Un­der­stand­ing Symbols

Jan 15, 2024, 9:21 PM
33 points
0 comments1 min readLW link

Live Sound: Big-O Improvements

jefftkJan 15, 2024, 7:50 PM
8 points
0 comments1 min readLW link
(www.jefftk.com)

In­ves­ti­gat­ing Bias Rep­re­sen­ta­tions in LLMs via Ac­ti­va­tion Steering

DawnLuJan 15, 2024, 7:39 PM
29 points
4 comments5 min readLW link

Sparse MLP Distillation

slavachalnevJan 15, 2024, 7:39 PM
30 points
3 comments6 min readLW link

Re­view of Align­ment Plan Cri­tiques- De­cem­ber AI-Plans Cri­tique-a-Thon Re­sults

IknownothingJan 15, 2024, 7:37 PM
24 points
0 comments25 min readLW link
(aiplans.substack.com)

[Question] What does it look like for AI to sig­nifi­cantly im­prove hu­man co­or­di­na­tion, be­fore su­per­in­tel­li­gence?

Bird ConceptJan 15, 2024, 7:22 PM
22 points
2 comments1 min readLW link

Now Ac­cept­ing Player Ap­pli­ca­tions for Band of Blades

Joe RogeroJan 15, 2024, 5:58 PM
2 points
0 comments3 min readLW link

Three Types of Con­straints in the Space of Agents

Jan 15, 2024, 5:27 PM
26 points
3 comments17 min readLW link

The case for train­ing fron­tier AIs on Sume­rian-only corpus

Jan 15, 2024, 4:40 PM
130 points
16 comments3 min readLW link

How to Pro­mote More Pro­duc­tive Dialogue Out­side of LessWrong

sweenesmJan 15, 2024, 2:16 PM
18 points
4 comments2 min readLW link

[Question] Come and day­dream with me about sci­ence reform

TeaTieAndHatJan 15, 2024, 11:09 AM
9 points
1 comment1 min readLW link

AI do­ing philos­o­phy = AI gen­er­at­ing hands?

Wei DaiJan 15, 2024, 9:04 AM
46 points
23 commentsLW link

Even if we lose, we win

MorphismJan 15, 2024, 2:15 AM
24 points
17 comments4 min readLW link

De­tach­ment vs at­tach­ment [AI risk and men­tal health]

Neil Jan 15, 2024, 12:41 AM
15 points
4 comments3 min readLW link

Mak­ing up statis­tics to es­tab­lish pri­or­ity on Land Value Tax vs Earned In­come Tax Credit vs So­cial Me­dia Dy­namic Regulation

CanucklugJan 14, 2024, 11:57 PM
−5 points
2 comments7 min readLW link

Is the uni­verse all there is? ‘Ev­i­dence’ for ob­jects out­side the uni­verse...

JonathanHallJan 14, 2024, 11:56 PM
−4 points
27 comments11 min readLW link

[Question] What is the min­i­mum amount of time travel and re­sources needed to se­cure the fu­ture?

PerhapsJan 14, 2024, 10:01 PM
−3 points
5 comments1 min readLW link

Gothen­burg LW /​ ACX meetup

StefanJan 14, 2024, 9:21 PM
1 point
0 comments1 min readLW link

Gothen­burg LW /​ ACX meetup

StefanJan 14, 2024, 9:20 PM
1 point
1 comment1 min readLW link

D&D.Sci Hyper­sphere Anal­y­sis Part 2: Non­lin­ear Effects & Interactions

aphyerJan 14, 2024, 7:59 PM
24 points
0 comments7 min readLW link

Gen­der Exploration

sapphireJan 14, 2024, 6:57 PM
117 points
26 comments5 min readLW link
(open.substack.com)

List of pro­jects that seem im­pact­ful for AI Governance

Jan 14, 2024, 4:53 PM
14 points
0 comments13 min readLW link

The Leeroy Jenk­ins prin­ci­ple: How faulty AI could guaran­tee “warn­ing shots”

titotalJan 14, 2024, 3:03 PM
48 points
6 commentsLW link
(titotal.substack.com)

No­tice When Peo­ple Are Direc­tion­ally Correct

Chris_LeongJan 14, 2024, 2:12 PM
136 points
8 comments2 min readLW link

Cor­ro­sive Mnemonics

EpiritoJan 14, 2024, 12:44 PM
7 points
0 comments2 min readLW link

Against most, but not all, AI risk analogies

Matthew BarnettJan 14, 2024, 3:36 AM
63 points
41 comments7 min readLW link

Vote With Your Face

jefftkJan 14, 2024, 3:30 AM
11 points
0 comments1 min readLW link
(www.jefftk.com)

Case Stud­ies in Re­v­erse-Eng­ineer­ing Sparse Au­toen­coder Fea­tures by Us­ing MLP Linearization

Jan 14, 2024, 2:06 AM
24 points
0 comments42 min readLW link

D&D.Sci Hyper­sphere Anal­y­sis Part 1: Datafields & Pre­limi­nary Analysis

aphyerJan 13, 2024, 8:16 PM
29 points
1 comment5 min readLW link

Some ad­di­tional SAE thoughts

HoagyJan 13, 2024, 7:31 PM
31 points
4 comments13 min readLW link

(4 min read) An in­tu­itive ex­pla­na­tion of the AI in­fluence situation

trevorJan 13, 2024, 5:34 PM
12 points
26 comments4 min readLW link

AI #47: Meet the New Year

ZviJan 13, 2024, 4:20 PM
36 points
7 comments57 min readLW link
(thezvi.wordpress.com)

Take­aways from the NeurIPS 2023 Tro­jan De­tec­tion Competition

mikesJan 13, 2024, 12:35 PM
20 points
2 comments1 min readLW link
(confirmlabs.org)

[Question] Why do so many think de­cep­tion in AI is im­por­tant?

PrometheusJan 13, 2024, 8:14 AM
24 points
12 comments1 min readLW link

Elimi­nat­ing Cookie Ban­ners is Hard

jefftkJan 13, 2024, 3:00 AM
23 points
15 comments3 min readLW link
(www.jefftk.com)

In­tro­duc­ing Align­ment Stress-Test­ing at Anthropic

evhubJan 12, 2024, 11:51 PM
182 points
23 comments2 min readLW link

D&D.Sci(-fi): Coloniz­ing the SuperHyperSphere

abstractapplicJan 12, 2024, 11:36 PM
48 points
23 comments2 min readLW link

Com­mon­wealth Fu­sion Sys­tems is the Same Scale as OpenAI

Jeffrey HeningerJan 12, 2024, 9:43 PM
22 points
13 comments2 min readLW link

Through­put vs. Latency

Jan 12, 2024, 9:37 PM
29 points
2 comments13 min readLW link

Sleeper Agents: Train­ing De­cep­tive LLMs that Per­sist Through Safety Training

Jan 12, 2024, 7:51 PM
305 points
95 comments3 min readLW link
(arxiv.org)

METAPHILOSOPHY—A Philoso­phiz­ing through log­i­cal consequences

SeremoniaJan 12, 2024, 6:47 PM
−7 points
7 comments1 min readLW link

Ideal­ism, Real­is­tic & Pragmatic

SeremoniaJan 12, 2024, 6:16 PM
−7 points
3 comments1 min readLW link