AI Im­pacts Quar­terly Newslet­ter, Jan-Mar 2023

HarlanApr 17, 2023, 10:10 PM
5 points
0 comments3 min readLW link
(blog.aiimpacts.org)

[Question] How do you al­ign your emo­tions through up­dates and ex­is­ten­tial un­cer­tainty?

VojtaKovarikApr 17, 2023, 8:46 PM
4 points
10 comments1 min readLW link

AI Align­ment Re­search Eng­ineer Ac­cel­er­a­tor (ARENA): call for applicants

CallumMcDougallApr 17, 2023, 8:30 PM
100 points
9 comments7 min readLW link

AI policy ideas: Read­ing list

Zach Stein-PerlmanApr 17, 2023, 7:00 PM
24 points
7 comments4 min readLW link

NYT: The Sur­pris­ing Thing A.I. Eng­ineers Will Tell You if You Let Them

SodiumApr 17, 2023, 6:59 PM
11 points
2 comments1 min readLW link
(www.nytimes.com)

But why would the AI kill us?

So8resApr 17, 2023, 6:42 PM
140 points
96 comments2 min readLW link

Sama Says the Age of Gi­ant AI Models is Already Over

AlgonApr 17, 2023, 6:36 PM
49 points
12 comments1 min readLW link
(www.wired.com)

Meetup Tip: Con­ver­sa­tion Starters

ScrewtapeApr 17, 2023, 6:25 PM
20 points
1 comment3 min readLW link

Cri­tiques of promi­nent AI safety labs: Red­wood Research

Omega.Apr 17, 2023, 6:20 PM
4 points
0 comments22 min readLW link
(forum.effectivealtruism.org)

How Large Lan­guage Models Nuke our Naive No­tions of Truth and Reality

Sean LeeApr 17, 2023, 6:08 PM
0 points
23 comments11 min readLW link

An al­ter­na­tive of PPO to­wards alignment

ml hkustApr 17, 2023, 5:58 PM
2 points
2 comments4 min readLW link

What I learned at the AI Safety Europe Retreat

skaisgApr 17, 2023, 5:40 PM
28 points
0 comments10 min readLW link
(skaisg.eu)

What is your timelines for ADI (ar­tifi­cial dis­em­pow­er­ing in­tel­li­gence)?

Christopher KingApr 17, 2023, 5:01 PM
3 points
3 comments2 min readLW link

[Question] Can we get around Godel’s In­com­plete­ness the­o­rems and Tur­ing un­de­cid­able prob­lems via in­finite com­put­ers?

Noosphere89Apr 17, 2023, 3:14 PM
−11 points
12 comments1 min readLW link

La Crosse, WI Ra­tion­al­ity Meetup

Daniel UebeleApr 17, 2023, 3:13 PM
1 point
0 comments1 min readLW link

Slow­ing AI: Foundations

Zach Stein-PerlmanApr 17, 2023, 2:30 PM
45 points
11 comments17 min readLW link

Slow­ing AI: Read­ing list

Zach Stein-PerlmanApr 17, 2023, 2:30 PM
47 points
3 comments4 min readLW link

Good­hart’s Law in­side the hu­man mind

Kaj_SotalaApr 17, 2023, 1:48 PM
125 points
13 comments16 min readLW link

Pre­dic­tion: any un­con­trol­lable AI will turn earth into a gi­ant computer

Karl von WendtApr 17, 2023, 12:30 PM
11 points
8 comments3 min readLW link

Au­toBound on neu­ral net­work can achieve OOMs lower train­ing loss

Maybe_aApr 17, 2023, 5:20 AM
10 points
9 comments1 min readLW link
(ai.googleblog.com)

Mak­ing Book­ing.Com less out to get you

ElizabethApr 17, 2023, 4:04 AM
21 points
0 comments1 min readLW link
(www.alexcharlton.co)

grey goo is unlikely

bhauthApr 17, 2023, 1:59 AM
156 points
123 comments9 min readLW link2 reviews
(bhauth.com)

AGI Clinics: A Safe Haven for Hu­man­ity’s First En­coun­ters with Superintelligence

portr.Apr 17, 2023, 1:52 AM
−5 points
1 comment1 min readLW link

Sum­maries of top fo­rum posts (27th March to 16th April)

Zoe WilliamsApr 17, 2023, 12:28 AM
14 points
1 commentLW link

AI Takeover Sce­nario with Scaled LLMs

simeon_cApr 16, 2023, 11:28 PM
42 points
15 comments8 min readLW link

My ex­pe­rience get­ting fund­ing for my biolog­i­cal research

MetacelsusApr 16, 2023, 10:53 PM
78 points
10 comments5 min readLW link
(denovo.substack.com)

Top les­son from GPT: we will prob­a­bly de­stroy hu­man­ity “for the lulz” as soon as we are able.

ShmiApr 16, 2023, 8:27 PM
63 points
28 comments1 min readLW link

On ur­gency, pri­or­ity and col­lec­tive re­ac­tion to AI-Risks: Part I

DenreikApr 16, 2023, 7:14 PM
−10 points
15 comments5 min readLW link

Effi­cient Learn­ing: Memorization

Alvin ÅnestrandApr 16, 2023, 5:58 PM
4 points
2 comments5 min readLW link
(forum.effectivealtruism.org)

Mechanis­ti­cally in­ter­pret­ing time in GPT-2 small

Apr 16, 2023, 5:57 PM
68 points
6 comments21 min readLW link

La Crosse, WI Ra­tion­al­ity Meetup

Daniel UebeleApr 16, 2023, 5:33 PM
1 point
0 comments1 min readLW link

The Soul of the Writer (on LLMs, the psy­chol­ogy of writ­ers, and the na­ture of in­tel­li­gence)

rogersbaconApr 16, 2023, 4:02 PM
11 points
1 comment3 min readLW link
(www.secretorum.life)

Pos­si­bi­liz­ing vs. actualizing

TsviBTApr 16, 2023, 3:55 PM
31 points
2 comments5 min readLW link

Hu­man Ex­tinc­tion by AI through eco­nomic power

ChristianKlApr 16, 2023, 12:15 PM
8 points
1 comment8 min readLW link

Bit Flip

Charlie SandersApr 16, 2023, 7:30 AM
−2 points
11 comments11 min readLW link

Dou­ble-nega­tion as framing

Stuart JohnsonApr 16, 2023, 6:59 AM
25 points
9 comments6 min readLW link

[Link/​cross­post] [US] NTIA: AI Ac­countabil­ity Policy Re­quest for Comment

Kyle J. LuccheseApr 16, 2023, 6:57 AM
8 points
0 comments1 min readLW link
(forum.effectivealtruism.org)

[Question] Who is test­ing AI Safety pub­lic out­reach mes­sag­ing?

yanni kyriacosApr 16, 2023, 6:57 AM
13 points
2 comments1 min readLW link

Fea­tures of Emacs that I only re­cently discovered

EmacsScrubApr 16, 2023, 6:57 AM
12 points
5 comments3 min readLW link

ACX meetup in Prague (16th of May)

Jiří NádvorníkApr 16, 2023, 6:25 AM
4 points
0 comments1 min readLW link

Smar­tyHead­erCode: anoma­lous to­kens for GPT3.5 and GPT-4

AdamYedidiaApr 15, 2023, 10:35 PM
71 points
18 comments6 min readLW link

Open-source LLMs may prove Bostrom’s vuln­er­a­ble world hypothesis

Roope AhvenharjuApr 15, 2023, 7:16 PM
1 point
1 comment1 min readLW link

[linkpost] Elon Musk plans AI start-up to ri­val OpenAI

HatfieldApr 15, 2023, 7:06 PM
11 points
11 comments1 min readLW link
(www.ft.com)

FLI re­port: Poli­cy­mak­ing in the Pause

Zach Stein-PerlmanApr 15, 2023, 5:01 PM
15 points
3 comments1 min readLW link
(futureoflife.org)

Reflec­tive jour­nal en­tries us­ing GPT-4 and Ob­sidian that de­mand less willpower.

Solenoid_EntityApr 15, 2023, 12:45 PM
56 points
24 comments7 min readLW link

An ex­am­ple ele­va­tor pitch for AI doom

laserficheApr 15, 2023, 12:29 PM
2 points
5 comments1 min readLW link

AI as Con­tact with our Col­lec­tive Unconscious

Scott BroockApr 15, 2023, 2:11 AM
−4 points
6 comments4 min readLW link

The Truth About False

Thoth HermesApr 15, 2023, 1:01 AM
−21 points
4 comments17 min readLW link
(thothhermes.substack.com)

The ‘ pe­ter­todd’ phenomenon

mwatkinsApr 15, 2023, 12:59 AM
192 points
50 comments38 min readLW link1 review

[Question] Con­cave Utility Question

Scott GarrabrantApr 15, 2023, 12:14 AM
55 points
36 comments2 min readLW link