Maybe I was too harsh on deep learning theory (three days ago)

LawrenceC30 Apr 2026 6:57 UTC

37 points

0 comments4 min readLW link

Red vs blue: The parable of the feud within a feud

Joe Rogero30 Apr 2026 4:01 UTC

9 points

3 comments5 min readLW link

(subatomicarticles.com)

Scaffolding vs Reinforcement Finetuning for AI Forecasting

Ram Potham30 Apr 2026 2:51 UTC

10 points

0 comments4 min readLW link

No Strong Orthogonality From Selection Pressure

lumpenspace30 Apr 2026 1:56 UTC

48 points

6 comments10 min readLW link

Computation in Superposition: Two Handcrafted Models

RGRGRG and Kyle Ray

30 Apr 2026 0:58 UTC

16 points

0 comments7 min readLW link

Research Sabotage in ML Codebases

egan, Vivek Hebbar and Julian Stastny

30 Apr 2026 0:26 UTC

23 points

0 comments6 min readLW link

The fall of the theorem economy (David Bessis)

Caleb Biddulph29 Apr 2026 19:35 UTC

21 points

7 comments4 min readLW link

(davidbessis.substack.com)

Probe-Based Data Attribution: Surfacing and Mitigating Undesirable Behaviors in LLM Post-Training

Santiago Aranguri and frankyaoxiao

29 Apr 2026 19:30 UTC

12 points

0 comments13 min readLW link

Book review: The Infinity Machine

PeterMcCluskey29 Apr 2026 18:59 UTC

24 points

1 comment6 min readLW link

Poisoning Fine-tuning Datasets of Constitutional Classifiers

Chase Bowers and Fabien Roger

29 Apr 2026 17:04 UTC

26 points

2 comments11 min readLW link

(alignment.anthropic.com)

AGI is Probably Inevitable: A Model of Societal Ruptures

Mira Kennard29 Apr 2026 16:00 UTC

4 points

0 comments5 min readLW link

Final research agenda #2: first sketch of a plan

Mitchell_Porter29 Apr 2026 15:19 UTC

21 points

0 comments4 min readLW link

The Enneagram is a Useful Fake Framework

Gordon Seidoh Worley29 Apr 2026 14:30 UTC

7 points

0 comments3 min readLW link

(www.uncertainupdates.com)

The Most Important Charts In The World

Zvi29 Apr 2026 14:10 UTC

51 points

1 comment2 min readLW link

(thezvi.wordpress.com)

Pears

TylerH29 Apr 2026 13:17 UTC

−1 points

1 comment4 min readLW link

Learning zero, and what SLT gets wrong about it

Dmitry Vaintrob29 Apr 2026 6:41 UTC

34 points

2 comments13 min readLW link

Are LLMs not getting better?

kqr29 Apr 2026 6:27 UTC

23 points

4 comments2 min readLW link

llm assistant personas seem increasingly incoherent (some subjective observations)

nostalgebraist29 Apr 2026 3:53 UTC

156 points

46 comments9 min readLW link

Not a Paper: “Frontier Lab CEOs are Capable of In-Context Scheming”

LawrenceC29 Apr 2026 3:00 UTC

164 points

4 comments7 min readLW link

Notes on Transformer Consciousness

slavachalnev29 Apr 2026 0:00 UTC

30 points

1 comment2 min readLW link