Mechanistic estimation for expectations of random products

Jacob_Hilton, George Robinson, Eric Neyman, paulfchristiano, Mikewins, Victor Lecomte, Wilson Wu and Gabriel Wu

15 May 2026 16:50 UTC

14 points

0 comments5 min readLW link

(www.alignment.org)

The safe-to-dangerous shift is a fundamental problem for eval realism; but also for measuring awareness

Charlie Griffin and Patrick Leask

14 May 2026 17:05 UTC

59 points

1 comment3 min readLW link

Empowerment, corrigibility, etc. are simple abstractions (of a messed-up ontology)

Steven Byrnes11 May 2026 17:48 UTC

113 points

50 comments16 min readLW link

Clarifying the role of the behavioral selection model

Alex Mallen10 May 2026 19:41 UTC

17 points

0 comments4 min readLW link

Natural Language Autoencoders Produce Unsupervised Explanations of LLM Activations

Subhash Kantamneni, kitft, Euan Ong and Sam Marks

7 May 2026 20:21 UTC

211 points

29 comments8 min readLW link

Mechanistic estimation for wide random MLPs

Jacob_Hilton7 May 2026 16:20 UTC

75 points

5 comments5 min readLW link

(www.alignment.org)

[Linkpost] Interpreting Language Model Parameters

Lucius Bushnaq, Dan Braun, Oliver Clive-Griffin, Bart Bussmann, Nathan Hu, mivanitskiy, Linda Linsefors and Lee Sharkey

5 May 2026 17:37 UTC

159 points

2 comments2 min readLW link

(www.goodfire.ai)

Motivated reasoning, confirmation bias, and AI risk theory

Seth Herd5 May 2026 15:56 UTC

49 points

13 comments41 min readLW link

Exploration Hacking: Can LLMs Learn to Resist RL Training?

Eyon Jang, Joschka Braun, Damon Falck and David Lindner

1 May 2026 20:54 UTC

23 points

0 comments8 min readLW link

Risk from fitness-seeking AIs: mechanisms and mitigations

Alex Mallen1 May 2026 17:42 UTC

99 points

0 comments32 min readLW link

Research Sabotage in ML Codebases

egan, Vivek Hebbar and Julian Stastny

30 Apr 2026 0:26 UTC

62 points

3 comments6 min readLW link

Recursive forecasting: Eliciting long-term forecasts from myopic fitness-seekers

Jozdien and Alex Mallen

28 Apr 2026 18:00 UTC

55 points

2 comments7 min readLW link

Sleeper Agent Backdoor Results Are Messy

SebastianP, Alek Westover, Dylan Xu, Vivek Hebbar and Julian Stastny

28 Apr 2026 1:55 UTC

81 points

4 comments7 min readLW link

Language models know what matters and the foundations of ethics better than you

Michele Campolo27 Apr 2026 14:00 UTC

6 points

2 comments90 min readLW link

From nothing to important actions: agents that act morally

Michele Campolo27 Apr 2026 13:59 UTC

1 point

0 comments22 min readLW link

The other paper that killed deep learning theory

LawrenceC27 Apr 2026 6:57 UTC

83 points

6 comments8 min readLW link

The paper that killed deep learning theory

LawrenceC26 Apr 2026 6:55 UTC

186 points

10 comments6 min readLW link

Quick Paper Review: “There Will Be a Scientific Theory of Deep Learning”

LawrenceC25 Apr 2026 6:55 UTC

108 points

3 comments6 min readLW link

A “Lay” Introduction to “On the Complexity of Neural Computation in Superposition”

LawrenceC22 Apr 2026 2:26 UTC

33 points

0 comments5 min readLW link

Preventing extinction from ASI on a $50M yearly budget

Andrea_Miotti, Alex Amadori and Gabriel Alfour

21 Apr 2026 16:39 UTC

212 points

54 comments21 min readLW link

(controlai.com)