Ex­per­i­ment Idea: RL Agents Evad­ing Learned Shutdownability

Leon LangJan 16, 2023, 10:46 PM
31 points
7 comments17 min readLW link
(docs.google.com)

Con­se­quen­tial­ists: One-Way Pat­tern Traps

David UdellJan 16, 2023, 8:48 PM
59 points
3 comments14 min readLW link

Book Re­view: Wor­lds of Flow

rememberJan 16, 2023, 8:17 PM
83 points
3 comments9 min readLW link

For the Record: DL ∩ ASI = ∅

maximkazhenkovJan 16, 2023, 7:04 PM
13 points
13 comments2 min readLW link

[Question] What de­ter­mines fe­male ro­man­tic “mar­ket value”?

anon_girlJan 16, 2023, 6:45 PM
16 points
53 comments1 min readLW link

Sta­tus conscious

avantika.mehraJan 16, 2023, 5:44 PM
2 points
0 comments5 min readLW link

Con­fus­ing the ideal for the necessary

adamShimiJan 16, 2023, 5:29 PM
79 points
6 comments1 min readLW link
(epistemologicalvigilance.substack.com)

Tyler Cowen AMA on the Progress Forum

jasoncrawfordJan 16, 2023, 5:23 PM
19 points
0 comments1 min readLW link
(progressforum.org)

Reflec­tions on Trust­ing Trust & AI

Itay YonaJan 16, 2023, 6:36 AM
10 points
1 comment3 min readLW link
(mentaleap.ai)

Is “Earn­ing to Give” a Bad Frame­work?

clansJan 16, 2023, 5:35 AM
2 points
4 comments6 min readLW link
(locationtbd.home.blog)

Why you ask the sig­nifi­cance ques­tion why

SliderJan 16, 2023, 3:44 AM
6 points
0 comments1 min readLW link

In­vest­ment, Work, and Vi­sion: Who is re­spon­si­ble for cre­at­ing value?

SableJan 16, 2023, 1:57 AM
0 points
10 comments8 min readLW link
(affablyevil.substack.com)

Con­clu­sion and Bibliog­ra­phy for “Un­der­stand­ing the diffu­sion of large lan­guage mod­els”

Ben CottierJan 16, 2023, 1:46 AM
4 points
0 commentsLW link

Ques­tions for fur­ther in­ves­ti­ga­tion of AI diffusion

Ben CottierJan 16, 2023, 1:46 AM
4 points
0 commentsLW link

Im­pli­ca­tions of large lan­guage model diffu­sion for AI governance

Ben CottierJan 16, 2023, 1:45 AM
7 points
0 commentsLW link

Publi­ca­tion de­ci­sions for large lan­guage mod­els, and their impacts

Ben CottierJan 16, 2023, 1:44 AM
4 points
0 commentsLW link

Drivers of large lan­guage model diffu­sion: in­cre­men­tal re­search, pub­lic­ity, and cascades

Ben CottierJan 16, 2023, 1:44 AM
4 points
0 commentsLW link

The repli­ca­tion and em­u­la­tion of GPT-3

Ben CottierJan 16, 2023, 1:40 AM
4 points
0 commentsLW link

GPT-3-like mod­els are now much eas­ier to ac­cess and de­ploy than to develop

Ben CottierJan 16, 2023, 1:39 AM
12 points
3 commentsLW link

Back­ground for “Un­der­stand­ing the diffu­sion of large lan­guage mod­els”

Ben CottierJan 16, 2023, 1:38 AM
4 points
0 commentsLW link

Un­der­stand­ing the diffu­sion of large lan­guage mod­els: summary

Ben CottierJan 16, 2023, 1:37 AM
26 points
1 commentLW link

Spec­u­la­tion on Path-Depen­dance in Large Lan­guage Models.

NickyPJan 15, 2023, 8:42 PM
16 points
2 comments7 min readLW link

Un­der­speci­fi­ca­tion of Or­a­cle AI

Jan 15, 2023, 8:10 PM
30 points
12 comments19 min readLW link

[Question] How Does the Hu­man Brain Com­pare to Deep Learn­ing on Sam­ple Effi­ciency?

DragonGodJan 15, 2023, 7:49 PM
11 points
6 comments1 min readLW link

De­cep­tive failures short of full catas­tro­phe.

Alex Lawsen Jan 15, 2023, 7:28 PM
33 points
5 comments9 min readLW link

Non-di­rected con­cep­tual found­ing

TsviBTJan 15, 2023, 2:56 PM
12 points
3 comments1 min readLW link

Panop­ti­cons aren’t enough

Program DenJan 15, 2023, 12:55 PM
−10 points
7 comments1 min readLW link

[Question] Is this chat GPT rewrite of my post bet­ter?

Yair HalberstadtJan 15, 2023, 9:47 AM
2 points
5 comments1 min readLW link

A sim­ple pro­posal for pre­serv­ing free speech on twitter

Yair HalberstadtJan 15, 2023, 9:42 AM
−2 points
13 comments1 min readLW link

Core Con­cept Con­ver­sa­tion: What is tech­nol­ogy?

Adam ZernerJan 15, 2023, 9:40 AM
8 points
1 comment1 min readLW link

Lan­guage Ex Machina

janusJan 15, 2023, 9:19 AM
42 points
23 comments24 min readLW link
(generative.ink)

Core Con­cept Con­ver­sa­tion: What is wealth?

Adam ZernerJan 15, 2023, 9:07 AM
13 points
30 comments3 min readLW link

Core Con­cept Conversations

Adam ZernerJan 15, 2023, 7:17 AM
14 points
1 comment1 min readLW link

In­cen­tives con­sid­ered harmful

Ulisse MiniJan 15, 2023, 6:38 AM
6 points
0 comments1 min readLW link
(uli.rocks)

Con­sider pay­ing for liter­a­ture or book re­views us­ing boun­ties and dom­i­nant as­surance contracts

Arjun PanicksseryJan 15, 2023, 3:56 AM
57 points
7 comments2 min readLW link

Pod­cast with Divia Eden on op­er­ant conditioning

DanielFilanJan 15, 2023, 2:44 AM
14 points
0 comments1 min readLW link
(youtu.be)

We Need Holis­tic AI Macrostrategy

NickGabsJan 15, 2023, 2:13 AM
39 points
4 comments8 min readLW link

[Question] When to men­tion ir­rele­vant ac­cu­sa­tions?

philhJan 14, 2023, 9:58 PM
20 points
50 comments1 min readLW link

World-Model In­ter­pretabil­ity Is All We Need

Thane RuthenisJan 14, 2023, 7:37 PM
36 points
22 comments21 min readLW link

Cur­rent AI Models Seem Suffi­cient for Low-Risk, Benefi­cial AI

harsimonyJan 14, 2023, 6:55 PM
17 points
1 comment2 min readLW link

[Question] Ba­sic Ques­tion about LLMs: how do they know what task to perform

GarakJan 14, 2023, 1:13 PM
1 point
3 comments1 min readLW link

Aligned with what?

Program DenJan 14, 2023, 10:28 AM
3 points
41 comments1 min readLW link

Wok­ism, re­think­ing pri­ori­ties and the Bostrom case

Arturo MaciasJan 14, 2023, 2:27 AM
−25 points
2 comments4 min readLW link

A gen­eral com­ment on dis­cus­sions of ge­netic group differences

anonymous8101Jan 14, 2023, 2:11 AM
71 points
46 comments3 min readLW link

Ab­strac­tions as mor­phisms be­tween (co)algebras

Erik JennerJan 14, 2023, 1:51 AM
17 points
1 comment8 min readLW link

Con­crete Rea­sons for Hope about AI

Zac Hatfield-DoddsJan 14, 2023, 1:22 AM
94 points
13 comments1 min readLW link

Nega­tive Ex­per­tise

Jonas KgomoJan 14, 2023, 12:51 AM
4 points
0 comments1 min readLW link
(twitter.com)

Mid-At­lantic AI Align­ment Alli­ance Unconference

QuinnJan 13, 2023, 8:33 PM
7 points
2 comments1 min readLW link

Smal­lpox vac­cines are widely available, for now

David HornbeinJan 13, 2023, 8:02 PM
26 points
5 comments1 min readLW link

How does GPT-3 spend its 175B pa­ram­e­ters?

Robert_AIZIJan 13, 2023, 7:21 PM
41 points
14 comments6 min readLW link
(aizi.substack.com)