In­ter­pret­ing OpenAI’s Whisper

EllenaRSep 24, 2023, 5:53 PM
116 points
13 comments7 min readLW link

Con­tra­dic­tion Ap­peal Bias

onurSep 24, 2023, 5:03 PM
3 points
2 comments1 min readLW link

RAIN: Your Lan­guage Models Can Align Them­selves with­out Fine­tun­ing—Microsoft Re­search 2023 - Re­duces the ad­ver­sar­ial prompt at­tack suc­cess rate from 94% to 19%!

Singularian2501Sep 24, 2023, 4:48 PM
5 points
0 comments1 min readLW link

Honor Sys­tem for Vac­ci­na­tion?

jefftkSep 24, 2023, 11:50 AM
17 points
22 comments1 min readLW link
(www.jefftk.com)

Far-Fu­ture Com­mit­ments as a Policy Con­sen­sus Strategy

FCCCSep 24, 2023, 6:34 AM
7 points
40 comments1 min readLW link

Five ne­glected work ar­eas that could re­duce AI risk

Sep 24, 2023, 2:03 AM
17 points
5 comments9 min readLW link

[Question] Are the other Ra­tion­al­ity: A-Z se­quences com­ing out as books?

caffeinated_dissonanceSep 24, 2023, 12:38 AM
7 points
4 comments1 min readLW link

The Dick Kick’em Paradox

Augs SMSHacksSep 23, 2023, 10:22 PM
−5 points
21 comments1 min readLW link

I de­signed an AI safety course (for a philos­o­phy de­part­ment)

Eleni AngelouSep 23, 2023, 10:03 PM
37 points
15 comments2 min readLW link

Paper: LLMs trained on “A is B” fail to learn “B is A”

Sep 23, 2023, 7:55 PM
121 points
74 comments4 min readLW link
(arxiv.org)

Sparse Cod­ing, for Mechanis­tic In­ter­pretabil­ity and Ac­ti­va­tion Engineering

David UdellSep 23, 2023, 7:16 PM
42 points
7 comments34 min readLW link

[Question] Places to meet in­ter­est­ing mid­dle-aged men?

anon_girlSep 23, 2023, 7:06 PM
18 points
6 comments1 min readLW link

Tak­ing fea­tures out of su­per­po­si­tion with sparse au­toen­coders more quickly with in­formed initialization

Pierre PeignéSep 23, 2023, 4:21 PM
30 points
8 comments5 min readLW link

A quick re­mark on so-called “hal­lu­ci­na­tions” in LLMs and hu­mans

Bill BenzonSep 23, 2023, 12:17 PM
4 points
4 comments1 min readLW link

Hand-writ­ing MathML

jefftkSep 23, 2023, 11:20 AM
16 points
40 comments1 min readLW link
(www.jefftk.com)

Musk, Star­link, and Crimea

Nicholas / Heather KrossSep 23, 2023, 2:35 AM
−13 points
0 comments5 min readLW link

[Linkpost/​Video] All The Times We Nearly Blew Up The World

Jacob G-WSep 23, 2023, 1:18 AM
6 points
1 comment1 min readLW link
(www.youtube.com)

Luck based medicine: in­os­i­tol for anx­iety and brain fog

ElizabethSep 22, 2023, 8:10 PM
40 points
5 comments3 min readLW link
(acesounderglass.com)

If in­fluence func­tions are not ap­prox­i­mat­ing leave-one-out, how are they sup­posed to help?

Fabien RogerSep 22, 2023, 2:23 PM
66 points
5 comments3 min readLW link

Model­ing p(doom) with TrojanGDP

K. Liam SmithSep 22, 2023, 2:19 PM
−2 points
2 comments13 min readLW link

Let’s talk about Im­pos­tor syn­drome in AI safety

Igor IvanovSep 22, 2023, 1:51 PM
30 points
4 comments3 min readLW link

Fund Tran­sit With Development

jefftkSep 22, 2023, 11:10 AM
47 points
22 comments3 min readLW link
(www.jefftk.com)

Atoms to Agents Proto-Lectures

johnswentworthSep 22, 2023, 6:22 AM
96 points
14 comments2 min readLW link
(www.youtube.com)

Would You Work Harder In The Least Con­ve­nient Pos­si­ble World?

FirinnSep 22, 2023, 5:17 AM
99 points
100 comments9 min readLW link2 reviews

Con­tra Kevin Dorst’s Ra­tional Polarization

azsantoskSep 22, 2023, 4:28 AM
8 points
2 comments9 min readLW link

ACX Bos­ton—Petrov Day 2023

duck_masterSep 22, 2023, 1:13 AM
2 points
0 comments1 min readLW link

What so­cial sci­ence re­search do you want to see re­an­a­lyzed?

Michael WiebeSep 22, 2023, 12:03 AM
14 points
9 comments1 min readLW link

Im­mor­tal­ity or death by AGI

ImmortalityOrDeathByAGISep 21, 2023, 11:59 PM
47 points
30 comments4 min readLW link
(forum.effectivealtruism.org)

Neel Nanda on the Mechanis­tic In­ter­pretabil­ity Re­searcher Mindset

Michaël TrazziSep 21, 2023, 7:47 PM
37 points
1 comment3 min readLW link
(theinsideview.ai)

Re­quire AGI to be Explainable

PeterMcCluskeySep 21, 2023, 4:11 PM
5 points
0 comments6 min readLW link
(bayesianinvestor.com)

Up­date to “Dom­i­nant As­surance Con­tract Plat­form”

moyamoSep 21, 2023, 4:09 PM
32 points
1 comment1 min readLW link

Sparse Au­toen­coders: Fu­ture Work

Sep 21, 2023, 3:30 PM
35 points
5 comments6 min readLW link

Sparse Au­toen­coders Find Highly In­ter­pretable Direc­tions in Lan­guage Models

Sep 21, 2023, 3:30 PM
159 points
8 comments5 min readLW link

There should be more AI safety orgs

Marius HobbhahnSep 21, 2023, 2:53 PM
181 points
25 comments17 min readLW link

Ward 5: Jack Perenick and Naima Sait

jefftkSep 21, 2023, 1:00 PM
24 points
0 comments1 min readLW link
(www.jefftk.com)

AI #30: Dalle-3 and GPT-3.5-In­struct-Turbo

ZviSep 21, 2023, 12:00 PM
75 points
8 comments47 min readLW link
(thezvi.wordpress.com)

[Question] How are ra­tio­nal­ists or orgs blocked, that you can see?

Nathan YoungSep 21, 2023, 2:37 AM
7 points
2 comments1 min readLW link

Vi­sion Week­end US Edition

Allison DuettmannSep 20, 2023, 9:28 PM
4 points
0 comments1 min readLW link

Fore­sight Vi­sion Week­end Europe Edition

Allison DuettmannSep 20, 2023, 9:25 PM
3 points
0 comments1 min readLW link

Notes on ChatGPT’s “mem­ory” for strings and for events

Bill BenzonSep 20, 2023, 6:12 PM
3 points
0 comments10 min readLW link

Belief and the Truth

Sam I amSep 20, 2023, 5:38 PM
2 points
14 comments5 min readLW link
(open.substack.com)

Image Hi­jacks: Ad­ver­sar­ial Images can Con­trol Gen­er­a­tive Models at Runtime

Sep 20, 2023, 3:23 PM
58 points
9 comments1 min readLW link
(arxiv.org)

In­ter­pretabil­ity Ex­ter­nal­ities Case Study—Hun­gry Hun­gry Hippos

Magdalena WacheSep 20, 2023, 2:42 PM
64 points
22 comments2 min readLW link

An Ele­men­tary In­tro­duc­tion to In­fra-Bayesianism

CharlesRWSep 20, 2023, 2:29 PM
16 points
0 comments1 min readLW link

Weekly In­ci­dence In­clud­ing Delay

jefftkSep 20, 2023, 2:00 PM
11 points
0 comments2 min readLW link
(www.jefftk.com)

[Question] The stereo­type of male clas­si­cal mu­sic lovers be­ing gay

BB6Sep 20, 2023, 1:23 PM
11 points
6 comments1 min readLW link

Hous­ing Roundup #6

ZviSep 20, 2023, 1:10 PM
27 points
8 comments14 min readLW link
(thezvi.wordpress.com)

Care­less talk on US-China AI com­pe­ti­tion? (and crit­i­cism of CAIS cov­er­age)

Oliver SourbutSep 20, 2023, 12:46 PM
16 points
3 comments10 min readLW link3 reviews
(www.oliversourbut.net)

A New Bayesian De­ci­sion Theory

Pareto OptimalSep 20, 2023, 9:36 AM
−6 points
0 comments1 min readLW link
(paretooptimal.substack.com)

Protest against Meta’s ir­re­versible pro­lifer­a­tion (Sept 29, San Fran­cisco)

Holly_ElmoreSep 19, 2023, 11:40 PM
54 points
33 commentsLW link