You Can’t Iterate to Trustworthy AI Code Without Understanding

ronbodkin17 Aug 2026 23:00 UTC

8 points

0 comments9 min readLW link

What gives you away: how LLMs form opinions of you

Cat McGee17 Aug 2026 20:59 UTC

20 points

4 comments6 min readLW link

Evaluating Chain-of-Thought Monitorability is Still an Open Problem: Comments on OpenAI’s Monitorability Evals

Connor Dilgren and Sarah Wiegreffe

17 Aug 2026 19:45 UTC

15 points

1 comment18 min readLW link

Weird Re-Tokenization, Symmetries and Compression: Research Agenda

Xenomirant and Sami Wolf

17 Aug 2026 19:43 UTC

12 points

1 comment16 min readLW link

Fork Around and Find Out Part 3: Interpreting the knight auditor

dl2717 Aug 2026 18:11 UTC

15 points

0 comments5 min readLW link

Connect to your future selves

PatrickDFarley17 Aug 2026 14:57 UTC

10 points

0 comments11 min readLW link

Study Update: Does post-training quantization change welfare-relevant indicators in open-weight language models?

ashesfall16 Aug 2026 21:11 UTC

7 points

0 comments7 min readLW link

Q2.5 2026 Timelines Update: Uplift and Revenue

brendanhalstead, Daniel Kokotajlo and elifland

16 Aug 2026 19:00 UTC

54 points

15 comments11 min readLW link

(blog.aifutures.org)

Will There Be an AI Hegemon? A Mental Model for AI Power Concentration

simeon_c16 Aug 2026 18:41 UTC

21 points

6 comments3 min readLW link

(simeoncampos.substack.com)

The Doomsday Argument is Reasonable and Mostly Points to Longevity

Josh Snider16 Aug 2026 17:14 UTC

27 points

5 comments6 min readLW link

Three thoughts on civilisational handoff

Cleo Nardo16 Aug 2026 17:12 UTC

56 points

2 comments2 min readLW link

(clattubato.substack.com)

Should Less Wrong add subtitles?

Chris_Leong16 Aug 2026 9:05 UTC

29 points

8 comments1 min readLW link

Are questions allowed on LessWrong?

yatharth16 Aug 2026 7:04 UTC

13 points

7 comments2 min readLW link

Does DiffusionGemma do latent reasoning?

Jan Bauer and Neel Nanda

16 Aug 2026 4:22 UTC

36 points

0 comments9 min readLW link

Kimi likes causal decision theory more after RL in twin prisoner’s dilemmas

oakhu15 Aug 2026 22:31 UTC

104 points

41 comments6 min readLW link

What if Parameter Updates were Text?

DaemonicSigil15 Aug 2026 20:06 UTC

16 points

0 comments11 min readLW link

How To Catch a Distilled Model

Goutham Nalagatla and Carlos Guerrero Alvarez

15 Aug 2026 19:09 UTC

12 points

8 comments9 min readLW link

Learning new facts can change LLM behaviour

Richard Juggins15 Aug 2026 13:46 UTC

33 points

2 comments14 min readLW link

(www.workingthroughai.com)

Rerunning AI safety papers on every frontier release would be pretty easy and valuable

Zephaniah Roe, hersheys and yix

15 Aug 2026 5:14 UTC

105 points

7 comments4 min readLW link

(secondlookresearch.com)

Metaphilosophy II: Empirical Flywheels

interstice15 Aug 2026 2:08 UTC

18 points

7 comments5 min readLW link

(thermontology.com)