All 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 202220232024 2025

All Jan Feb Mar AprMayJun Jul Aug Sep Oct Nov Dec

All 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 171819 20 21 22 23 24 25 26 27 28 29 30 31

Let’s use AI to harden human defenses against AI manipulation

Tom DavidsonMay 17, 2023, 11:33 PM

35 points

17 votes

Overall karma indicates overall quality.

7 comments24 min readLW link

Improving the safety of AI evals

JustinShovelain and Elliot Mckernon

May 17, 2023, 10:24 PM

13 points

13 votes

Overall karma indicates overall quality.

7 comments7 min readLW link

Possible AI “Fire Alarms”

Chris_LeongMay 17, 2023, 9:56 PM

15 points

10 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

AI Alignment in The New Yorker

Eleni AngelouMay 17, 2023, 9:36 PM

8 points

4 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

(www.newyorker.com)

ACI #3: The Origin of Goals and Utility

Akira PyinyaMay 17, 2023, 8:47 PM

1 point

1 vote

Overall karma indicates overall quality.

0 comments6 min readLW link

What if they gave an Industrial Revolution and nobody came?

jasoncrawfordMay 17, 2023, 7:41 PM

94 points

49 votes

Overall karma indicates overall quality.

10 comments19 min readLW link

(rootsofprogress.org)

DCF Event Notes

jefftkMay 17, 2023, 5:30 PM

22 points

9 votes

Overall karma indicates overall quality.

7 comments3 min readLW link

(www.jefftk.com)

Hiatus: EA and LW post summaries

Zoe WilliamsMay 17, 2023, 5:17 PM

14 points

8 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

[Question] When should I close the fridge?

lemonhopeMay 17, 2023, 4:56 PM

11 points

8 votes

Overall karma indicates overall quality.

11 comments1 min readLW link

Play Regrantor: Move up to $250,000 to Your Top High-Impact Projects!

Dawn DrescherMay 17, 2023, 4:51 PM

26 points

7 votes

Overall karma indicates overall quality.

0 comments2 min readLW link

(impactmarkets.substack.com)

Eisenhower’s Atoms for Peace Speech

Orpheus16May 17, 2023, 4:10 PM

18 points

5 votes

Overall karma indicates overall quality.

3 comments11 min readLW link

(www.iaea.org)

Creating a self-referential system prompt for GPT-4

OzyrusMay 17, 2023, 2:13 PM

3 points

3 votes

Overall karma indicates overall quality.

1 comment3 min readLW link

GPT-4 implicitly values identity preservation: a study of LMCA identity management

OzyrusMay 17, 2023, 2:13 PM

21 points

12 votes

Overall karma indicates overall quality.

4 comments13 min readLW link

Some quotes from Tuesday’s Senate hearing on AI

Daniel_EthMay 17, 2023, 12:13 PM

66 points

28 votes

Overall karma indicates overall quality.

9 comments4 min readLW link

Why AGI systems will not be fanatical maximisers (unless trained by fanatical humans)

titotalMay 17, 2023, 11:58 AM

5 points

16 votes

Overall karma indicates overall quality.

3 comments15 min readLW link

Conflicts between emotional schemas often involve internal coercion

Richard_NgoMay 17, 2023, 10:02 AM

43 points

22 votes

Overall karma indicates overall quality.

4 comments4 min readLW link

[Question] Is there a ‘time series forecasting’ equivalent of AIXI?

Solenoid_EntityMay 17, 2023, 4:35 AM

12 points

3 votes

Overall karma indicates overall quality.

2 comments1 min readLW link

$300 for the best sci-fi prompt

RomanSMay 17, 2023, 4:23 AM

40 points

19 votes

Overall karma indicates overall quality.

30 comments2 min readLW link

New User’s Guide to LessWrong

RubyMay 17, 2023, 12:55 AM

148 points

94 votes

Overall karma indicates overall quality.

59 comments11 min readLW link 1 review

Are AIs like Animals? Perspectives and Strategies from Biology

Jackson EmanuelMay 16, 2023, 11:39 PM

1 point

1 vote

Overall karma indicates overall quality.

0 comments21 min readLW link

A Mechanistic Interpretability Analysis of a GridWorld Agent-Simulator (Part 1 of N)

Joseph BloomMay 16, 2023, 10:59 PM

36 points

14 votes

Overall karma indicates overall quality.

2 comments16 min readLW link

A TAI which kills all humans might also doom itself

Jeffrey HeningerMay 16, 2023, 10:36 PM

7 points

9 votes

Overall karma indicates overall quality.

3 comments3 min readLW link

Brief notes on the Senate hearing on AI oversight

DizietMay 16, 2023, 10:29 PM

77 points

37 votes

Overall karma indicates overall quality.

2 comments2 min readLW link

$500 Bounty/Prize Problem: Channel Capacity Using “Insensitive” Functions

johnswentworthMay 16, 2023, 9:31 PM

40 points

14 votes

Overall karma indicates overall quality.

11 comments2 min readLW link

Progress links and tweets, 2023-05-16

jasoncrawfordMay 16, 2023, 8:54 PM

14 points

5 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

(rootsofprogress.org)

AI Will Not Want to Self-Improve

petersalibMay 16, 2023, 8:53 PM

28 points

33 votes

Overall karma indicates overall quality.

24 comments20 min readLW link

Nice intro video to RSI

Nathan Helm-BurgerMay 16, 2023, 6:48 PM

12 points

5 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

(youtu.be)

[Interview w/ Zvi Mowshowitz] Should we halt progress in AI?

fowlertmMay 16, 2023, 6:12 PM

18 points

10 votes

Overall karma indicates overall quality.

2 comments3 min readLW link

AI Risk & Policy Forecasts from Metaculus & FLI’s AI Pathways Workshop

_will_May 16, 2023, 6:06 PM

11 points

7 votes

Overall karma indicates overall quality.

4 comments8 min readLW link

[Question] Why doesn’t the presence of log-loss for probabilistic models (e.g. sequence prediction) imply that any utility function capable of producing a “fairly capable” agent will have at least some non-negligible fraction of overlap with human values?

Thoth HermesMay 16, 2023, 6:02 PM

2 points

4 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

Decision Theory with the Magic Parts Highlighted

moridinamaelMay 16, 2023, 5:39 PM

175 points

90 votes

Overall karma indicates overall quality.

24 comments5 min readLW link

We learn long-lasting strategies to protect ourselves from danger and rejection

Richard_NgoMay 16, 2023, 4:36 PM

87 points

52 votes

Overall karma indicates overall quality.

5 comments5 min readLW link

Proposal: Align Systems Earlier In Training

OneManyNoneMay 16, 2023, 4:24 PM

18 points

8 votes

Overall karma indicates overall quality.

0 comments11 min readLW link

Procedural Executive Function, Part 2

DaystarEldMay 16, 2023, 4:22 PM

24 points

11 votes

Overall karma indicates overall quality.

0 comments18 min readLW link

(daystareld.com)

My current workflow to study the internal mechanisms of LLM

Yulu PiMay 16, 2023, 3:27 PM

4 points

4 votes

Overall karma indicates overall quality.

0 comments1 min readLW link

Proposal: we should start referring to the risk from unaligned AI as a type of accident risk

Christopher KingMay 16, 2023, 3:18 PM

22 points

15 votes

Overall karma indicates overall quality.

6 comments2 min readLW link

AI Safety Newsletter #6: Examples of AI safety progress, Yoshua Bengio proposes a ban on AI agents, and lessons from nuclear arms control

Dan H and Orpheus16

May 16, 2023, 3:14 PM

31 points

10 votes

Overall karma indicates overall quality.

0 comments6 min readLW link

(newsletter.safe.ai)

Lazy Baked Mac and Cheese

jefftkMay 16, 2023, 2:40 PM

18 points

7 votes

Overall karma indicates overall quality.

2 comments1 min readLW link

(www.jefftk.com)

Tyler Cowen’s challenge to develop an ‘actual mathematical model’ for AI X-Risk

Joe BrentonMay 16, 2023, 11:57 AM

6 points

9 votes

Overall karma indicates overall quality.

4 comments1 min readLW link

Evaluating Language Model Behaviours for Shutdown Avoidance in Textual Scenarios

Simon Lermen, Teun van der Weij and Leon Lang

May 16, 2023, 10:53 AM

26 points

13 votes

Overall karma indicates overall quality.

0 comments13 min readLW link

[Review] Two People Smoking Behind the Supermarket

lsusrMay 16, 2023, 7:25 AM

32 points

22 votes

Overall karma indicates overall quality.

1 comment1 min readLW link

Superposition and Dropout

Edoardo PonaMay 16, 2023, 7:24 AM

21 points

11 votes

Overall karma indicates overall quality.

5 comments6 min readLW link

[Question] What is the literature on long term water fasts?

lcMay 16, 2023, 3:23 AM

16 points

7 votes

Overall karma indicates overall quality.

4 comments1 min readLW link

Lessons learned from offering in-office nutritional testing

ElizabethMay 15, 2023, 11:20 PM

80 points

41 votes

Overall karma indicates overall quality.

11 comments14 min readLW link

(acesounderglass.com)

Judgments often smuggle in implicit standards

Richard_NgoMay 15, 2023, 6:50 PM

97 points

51 votes

Overall karma indicates overall quality.

4 comments3 min readLW link

Rational retirement plans

IkMay 15, 2023, 5:49 PM

5 points

15 votes

Overall karma indicates overall quality.

17 comments1 min readLW link

[Question] (Crosspost) Asking for online calls on AI s-risks discussions

jackchang110May 15, 2023, 5:42 PM

1 point

1 vote

Overall karma indicates overall quality.

0 comments1 min readLW link

(forum.effectivealtruism.org)

Simple experiments with deceptive alignment

Andreas_MoeMay 15, 2023, 5:41 PM

7 points

6 votes

Overall karma indicates overall quality.

0 comments4 min readLW link

Some Summaries of Agent Foundations Work

mattmacdermottMay 15, 2023, 4:09 PM

62 points

30 votes

Overall karma indicates overall quality.

1 comment13 min readLW link

Facebook Increased Visibility

jefftkMay 15, 2023, 3:40 PM

15 points

10 votes

Overall karma indicates overall quality.

1 comment1 min readLW link

(www.jefftk.com)

Keyboard shortcuts

Keys shown in yellow (e.g., ]) are accesskeys, and require a browser-specific modifier key (or keys).

Keys shown in grey (e.g., ?) do not require any modifier keys.

General
? Show keyboard shortcuts
Esc Hide keyboard shortcuts

Site navigation
h Go to Home (a.k.a. “Frontpage”) view
f Go to Featured (a.k.a. “Curated”) view
a Go to All (a.k.a. “Community”) view
m Go to Meta view
v Go to Tags view
c Go to Recent Comments view
r Go to Archive view
q Go to Sequences view
t Go to About page
u Go to User or Login page
o Go to Inbox page

Page navigation
, Jump up to top of page
. Jump down to bottom of page
/ Jump to top of comments section
s Search

Page actions
n New post or comment
e Edit current post

Post/comment list views
. Focus next entry in list
, Focus previous entry in list
; Cycle between links in focused entry
Enter Go to currently focused entry
Esc Unfocus currently focused entry
] Go to next page
[ Go to previous page
\ Go to first page
e Edit currently focused post

Editor
k Bold text
i Italic text
l Insert hyperlink
q Blockquote text

Appearance
= Increase text size
- Decrease text size
0 Reset to default text size
′ Cycle through content width settings
1 Switch to default theme [A]
2 Switch to dark theme [B]
3 Switch to grey theme [C]
4 Switch to ultramodern theme [D]
5 Switch to simple theme [E]
6 Switch to brutalist theme [F]
7 Switch to ReadTheSequences theme [G]
8 Switch to classic Less Wrong theme [H]
9 Switch to modern Less Wrong theme [I]
; Open theme tweaker
Enter Save changes and close theme tweaker
Esc Close theme tweaker (without saving)

Slide shows
l Start/resume slideshow
Esc Exit slideshow
→↓ Next slide
←↑ Previous slide
Space Reset slide zoom

Miscellaneous
x Switch to next view on user page
z Switch to previous view on user page
` Toggle compact comment list view
g Toggle anti-kibitzer