27 Oct 2023 23:56 UTC

44 points

12 comments9 min readLW link

[Question] Do you believe “E=mc^2” is a correct and/or useful equation, and, whether yes or no, precisely what are your reasons for holding this belief (with such a degree of confidence)?

l8c27 Oct 2023 22:46 UTC

10 points

14 comments1 min readLW link

Value systematization: how values become coherent (and misaligned)

Richard_Ngo27 Oct 2023 19:06 UTC

102 points

48 comments13 min readLW link

Techno-humanism is techno-optimism for the 21st century

Richard_Ngo27 Oct 2023 18:37 UTC

88 points

5 comments14 min readLW link

(www.mindthefuture.info)

Sanctuary for Humans

Nikola Jurkovic27 Oct 2023 18:08 UTC

21 points

9 comments1 min readLW link

Wireheading and misalignment by composition on NetHack

pierlucadoro27 Oct 2023 17:43 UTC

34 points

4 comments4 min readLW link

We’re Not Ready: thoughts on “pausing” and responsible scaling policies

HoldenKarnofsky27 Oct 2023 15:19 UTC

200 points

33 comments8 min readLW link

Aspiration-based Q-Learning

Clément Dumas and Jobst Heitzig

27 Oct 2023 14:42 UTC

38 points

5 comments11 min readLW link

Linkpost: Rishi Sunak’s Speech on AI (26th October)

bideup27 Oct 2023 11:57 UTC

85 points

8 comments7 min readLW link

(www.gov.uk)

ASPR & WARP: Rationality Camps for Teens in Taiwan and Oxford

Anna Gajdova27 Oct 2023 8:40 UTC

18 points

0 comments1 min readLW link

[Question] To what extent is the UK Government’s recent AI Safety push entirely due to Rishi Sunak?

Stephen Fowler27 Oct 2023 3:29 UTC

23 points

4 comments1 min readLW link

Bayesian Punishment

Rob Lucas27 Oct 2023 3:24 UTC

1 point

1 comment6 min readLW link

Online Dialogues Party — Sunday 5th November

Ben Pace27 Oct 2023 2:41 UTC

28 points

1 comment1 min readLW link

OpenAI’s new Preparedness team is hiring

leopold26 Oct 2023 20:42 UTC

60 points

2 comments1 min readLW link

Fake Deeply

Zack_M_Davis26 Oct 2023 19:55 UTC

33 points

7 comments1 min readLW link

(unremediatedgender.space)

Symbol/Referent Confusions in Language Model Alignment Experiments

johnswentworth26 Oct 2023 19:49 UTC

94 points

44 comments6 min readLW link

Unsupervised Methods for Concept Discovery in AlphaZero

aogara26 Oct 2023 19:05 UTC

9 points

0 comments1 min readLW link

(arxiv.org)

[Question] Nonlinear limitations of ReLUs

magfrump26 Oct 2023 18:51 UTC

13 points

1 comment1 min readLW link

AI Alignment Problem: Requirement not optional (A Critical Analysis through Mass Effect Trilogy)

TAWSIF AHMED26 Oct 2023 18:02 UTC

−9 points

0 comments4 min readLW link

[Thought Experiment] Tomorrow’s Echo—The future of synthetic companionship.

Vimal Naran26 Oct 2023 17:54 UTC

−7 points

2 comments2 min readLW link

Disagreements over the prioritization of existential risk from AI

Olivier Coutu26 Oct 2023 17:54 UTC

10 points

0 comments6 min readLW link

[Question] What if AGI had its own universe to maybe wreck?

mseale26 Oct 2023 17:49 UTC

−1 points

2 comments1 min readLW link

Changing Contra Dialects

jefftk26 Oct 2023 17:30 UTC

25 points

2 comments1 min readLW link

(www.jefftk.com)

5 psychological reasons for dismissing x-risks from AGI

Igor Ivanov26 Oct 2023 17:21 UTC

24 points

6 comments4 min readLW link

5. Risks from preventing legitimate value change (value collapse)

Nora_Ammann26 Oct 2023 14:38 UTC

13 points

1 comment9 min readLW link

4. Risks from causing illegitimate value change (performative predictors)

Nora_Ammann26 Oct 2023 14:38 UTC

8 points

3 comments5 min readLW link

3. Premise three & Conclusion: AI systems can affect value change trajectories & the Value Change Problem

Nora_Ammann26 Oct 2023 14:38 UTC

28 points

4 comments4 min readLW link

2. Premise two: Some cases of value change are (il)legitimate

Nora_Ammann26 Oct 2023 14:36 UTC

24 points

7 comments6 min readLW link

1. Premise one: Values are malleable

Nora_Ammann26 Oct 2023 14:36 UTC

21 points

1 comment15 min readLW link

0. The Value Change Problem: introduction, overview and motivations

Nora_Ammann26 Oct 2023 14:36 UTC

32 points

0 comments5 min readLW link

EPUBs of MIRI Blog Archives and selected LW Sequences

mesaoptimizer26 Oct 2023 14:17 UTC

44 points

6 comments1 min readLW link

(git.sr.ht)

UK Government publishes “Frontier AI: capabilities and risks” Discussion Paper

A.H.26 Oct 2023 13:55 UTC

5 points

0 comments2 min readLW link

(www.gov.uk)

AI #35: Responsible Scaling Policies

Zvi26 Oct 2023 13:30 UTC

66 points

10 comments55 min readLW link

(thezvi.wordpress.com)

RA Bounty: Looking for feedback on screenplay about AI Risk

Writer26 Oct 2023 13:23 UTC

30 points

6 comments1 min readLW link

Sensor Exposure can Compromise the Human Brain in the 2020s

trevor26 Oct 2023 3:31 UTC

17 points

6 comments10 min readLW link

Notes on “How do we become confident in the safety of a machine learning system?”

RohanS26 Oct 2023 3:13 UTC

4 points

0 comments13 min readLW link

Apply to the Constellation Visiting Researcher Program and Astra Fellowship, in Berkeley this Winter

Nate Thomas26 Oct 2023 3:07 UTC

42 points

10 comments1 min readLW link

CHAI internship applications are open (due Nov 13)

Erik Jenner26 Oct 2023 0:53 UTC

34 points

0 comments3 min readLW link

Architects of Our Own Demise: We Should Stop Developing AI Carelessly

Roko26 Oct 2023 0:36 UTC

170 points

75 comments3 min readLW link

EA Infrastructure Fund: June 2023 grant recommendations

Linch26 Oct 2023 0:35 UTC

21 points

0 comments1 min readLW link

Responsible Scaling Policies Are Risk Management Done Wrong

simeon_c25 Oct 2023 23:46 UTC

122 points

35 comments22 min readLW link 1 review

(www.navigatingrisks.ai)

AI as a science, and three obstacles to alignment strategies

So8res25 Oct 2023 21:00 UTC

185 points

80 comments11 min readLW link

My hopes for alignment: Singular learning theory and whole brain emulation

Garrett Baker25 Oct 2023 18:31 UTC

61 points

5 comments12 min readLW link

[Question] Lying to chess players for alignment

Zane25 Oct 2023 17:47 UTC

96 points

54 comments1 min readLW link

Anthropic, Google, Microsoft & OpenAI announce Executive Director of the Frontier Model Forum & over $10 million for a new AI Safety Fund

Zach Stein-Perlman25 Oct 2023 15:20 UTC

31 points

8 comments4 min readLW link

(www.frontiermodelforum.org)

“The Economics of Time Travel”—call for reviewers (Seeds of Science)

rogersbacon25 Oct 2023 15:13 UTC

4 points

2 comments1 min readLW link

Compositional preference models for aligning LMs

Tomek Korbak25 Oct 2023 12:17 UTC

18 points

2 comments5 min readLW link

[Question] Should the US House of Representatives adopt rank choice voting for leadership positions?

jmh25 Oct 2023 11:16 UTC

16 points

6 comments1 min readLW link

Researchers believe they have found a way for artists to fight back against AI style capture

vernamcipher25 Oct 2023 10:54 UTC

3 points

1 comment1 min readLW link

(finance.yahoo.com)

Why We Disagree

zulupineapple25 Oct 2023 10:50 UTC

7 points

2 comments2 min readLW link