PODCAST

LessWrong (30+ Karma)

LessWrong

Audio narrations of LessWrong posts.

technology philosophy

All Episodes

32:07

“Sanity-checking “Incompressible Knowledge Probes””...

LessWrong (30+ Karma) ·

2026/05/01

Or, did a chief scientist of an AI assistant startup conclusively show that GPT-5.5 has 9.7 trillion parameters? Introduction Recently, a paper was circulated on Twitter claiming to have reverse engineered the parameter count of many frontier closed-source models including the newer GPT-5.5 (9.7 trillion parameters) and Claude Opus 4.7 (4.0 trillion parameters) as well as older models such as o1 (3.5T) and gpt-4o (720B). The paper, titled “Incompressible Knowledge Probes: Estimating Black-Box LLM

en-gb

3:32

“AI unemployment and AI extinction are often the...

LessWrong (30+ Karma) ·

2026/05/01

My sense is that people think of AI existential risk and AI unemployment as distinct issues.Some people are extremely concerned about extinction and perhaps even indifferent to total unemployment. Some people think of moderate AI unemployment as a realistic and concerning issue, and AI extinction as science fiction. I think of AI unemployment and AI extinction risk as basically the same issue, and in likely scenarios, happening together. At a very high level, I’d say the argument for human

en-gb

6:49

“AI risk was not invested by AI CEOs to hype their...

LessWrong (30+ Karma) ·

2026/05/01

I hear that many people believe that the idea of advanced AI threatening human existence was invented by AI CEOs to hype their products. I’ve even been condescendingly informed of this, as if I am the one at risk of naively accepting AI companies’ preferred narratives.If you are reading this, you are probably familiar enough with the decades-old AI safety community to know this isn’t true. But I don’t have a good direct way to reach the people who could use this information, and still I hate to

en-gb

10:29

“Cyborg evals” by Eye You, frmsaul

LessWrong (30+ Karma) ·

2026/04/30

The low-background steel problem Modern steel is slightly radioactive. We did a lot of atomic testing in the 40s and 50s, and now our atmosphere has some amount of radioactive particles, which make their way into steel during production. This is mostly fine, but some scientific instruments require steel that is not radioactive. In order to get such steel, people typically scavenge for pre-1950s steel (such as the sunken Imperial German Navy). Here's a photo of the SMS Hindenburg being salvaged. You

en-gb

54:29

“To what extent is Qwen3-32B predicting its persona?”...

LessWrong (30+ Karma) ·

2026/04/30

TL;DR We test to what extent Qwen3-32B behaves as though it is trying to predict what "Qwen3" would do. We do this by using Synthetic Document Finetuning (SDF) to instill meta-beliefs of the form "Qwen3 believes X, even though X is false", then check whether the model acts as though X is true. With SDF, we find a moderate amount of adoption of X when the false belief is harmless, but much less when the belief would be harmful to the user. When the meta-belief is instilled in-context instead, we see

en-gb

13:38

“Research Sabotage in ML Codebases” by egan

LessWrong (30+ Karma) ·

2026/04/30

One of the main hopes for AI safety is using AIs to automate AI safety research. However, if models are misaligned, then they may sabotage the safety research. For example, misaligned AIs may try to: Perform sloppy research in order to slow down the rate of research progressMake AI systems appear safer than they areTrain a successor model to be misaligned Whether we should worry about those things depends substantially on how hard it is to sabotage research in ways that are hard for reviewers to

en-gb

5:01

“Maybe I was too harsh on deep learning theory (three...

LessWrong (30+ Karma) ·

2026/04/30

A few days ago, I reviewed a paper titled “There Will Be a Scientific Theory of Deep Learning". In it, I expressed appreciation for the authors for writing the piece, but skepticism for stronger forms of their titular claims. Since then I’ve spoken with various past collaborators (via text and in person), and read or reread quite a few deep learning theory papers, including the bombshell Zhang et al. 2016 and Nagarajan et al. 2019 papers that I wrote about on LessWrong. And the thing is, parts of

en-gb

3:38

“Notes on Transformer Consciousness” by slavachalnev

LessWrong (30+ Karma) ·

2026/04/30

Assuming transformers can have conscious experience, what would that experience be like? Transformers[1] are a structured grid of layers and token positions and we can use this structure to reason about their internal experience. Epistemic status: very speculative. I've ordered this writeup approximately by how much I've thought it through and how much I believe it. Decode vs Prefill I claim that the experience of decode is identical to the experience of prefill. To see why this is the case,

en-gb

4:11

“On today’s panel with Bernie Sanders” by David Scott...

LessWrong (30+ Karma) ·

2026/04/30

It's sort of easy to forget how close Bernie Sanders was to becoming the most powerful person in the world. The world we live in feels so much not like that place. I’m in Washington DC for the next week, and I’ve just finished a public appearance with Senator Sanders (should I call him Bernie? Or Sanders? or…) You won’t often see me so dressed up and polished. But this is important! There are politicians who have principles and character, who really believe in doing what's right. I think you have

en-gb

20:00

“No Strong Orthogonality From Selection Pressure” by...

LessWrong (30+ Karma) ·

2026/04/30

TL;DR If everything goes according to plan, by the end of this post we should have separated three claims that are too often bundled together: Intelligence does not imply human morality.Weird minds are possible. 3.A reflective, recursively improving intelligence should be expected to remain bound to a semantically thin “terminal goal” that emerged during training. I accept the first two. I am arguing against the third. So: I am not making the case that sufficiently intelligent systems automatically

en-gb

22:53

“Learning zero, and what SLT gets wrong about it” by...

LessWrong (30+ Karma) ·

2026/04/30

This is a first in a pair of posts I'm hoping to write about Singular Learning Theory (SLT) and singularities as a model of data degeneracy. If I get to it, the second post is going to be more general-audience; this one is more technical. Introduction To me, SLT is an important source of toy models which point at an interesting class of new statistical phenomena in learning. It is also a valuable correction to an older and (at this point) largely-defunct story of learning being fully controlled by

en-gb

22:12

“LLM Style Slop is Absolutely Everywhere” by silentbob

LessWrong (30+ Karma) ·

2026/04/29

Epistemic status: something of a rant. Is not meant to make claims about the general capabilities (or lack thereof) of LLMs (beyond their prose), but observations about how society seems to use them excessively without being perfectly candid about it. Nanobanana's take on the situation Over the past few months, it has become a bit of a running gag that every evening, I inform my girlfriend of the multiple cases of LLM slop I encountered in the wild throughout the day[1]. And pretty much every

en-gb

7:47

“Goblin Mode, 24 Hours Later” by Dylan Bowman

LessWrong (30+ Karma) ·

2026/04/29

Yesterday, Twitter user arb8020 posted this: It went semi-viral within AI Twitter and users began experimenting with "goblin mode" and hypothesizing about the source of the bizarre behavior. LM Arena provided evidence for the phenomenon from their traffic: "It's true. Here's a plot of GPT models and their usage of 'goblin', 'gremlin', 'troll', etc over time. There's no anti-gremlin system instruction on our side, we get to see GPT-5.5 run free." — arena Some hypotheses about what causes this: "My

en-gb

2:40

“Let Kids Keep More Productivity Gains” by jefftk

LessWrong (30+ Karma) ·

2026/04/29

While I was traveling Julia asked me: why is Anna saying her fiddle practice is only two minutes? In this case, two minutes was the right amount of time! Anna (10y) and I had been fighting a lot about practice. She'd complain, slump, stop repeatedly to make adjustments, and generally be miserable. I'd often have to pull out "if you want to keep taking fiddle lessons you have to practice": she loves her teacher and is very motivated by the prospect of being good at fiddle. Still, it would

en-gb

15:47

“llm assistant personas seem increasingly incoherent...

LessWrong (30+ Karma) ·

2026/04/29

(This was originally going to be a "quick take" but then it got a bit long. Just FYI.) There's this weird trend I perceive with the personas of LLM assistants over time. It feels like they're getting less "coherent" in a certain sense, even as the models get more capable. When I read samples from older chat-tuned models, it's striking how "mode-collapsed" they feel relative to recent models like Claude Opus 4.6 or GPT-5.4.[1] This is most straightforwardly obvious when it comes to textual style and

en-gb

14:51

“Not a Paper: “Frontier Lab CEOs are Capable of...

LessWrong (30+ Karma) ·

2026/04/29

(Fragments from a research paper that will never be written) Extended Abstract. The frontier AI developers are becoming increasingly powerful and wealthy, significantly increasing their potential for risks. One concern is that of executive misalignment: when the CEO has different incentives and goals than that of the board of directors, or of humanity as a whole. Our work proposes three different threat models, under which executive misalignment can lead to concrete harm. We perform two

en-gb

19:15

“The Problem in the “Nerd Sniping” xkcd Comic” by...

LessWrong (30+ Karma) ·

2026/04/29

A few days ago I saw this comic reposted, and I thought: wait! Unlike every prior time I have seen this comic, I actually know how to solve this now! One thing which I often find really cool, and which I don’t think comes up a lot in most people's mathematical education, is when you can learn something about some thing non-random by analysing a random process. So, let me show you a way to find out the resistance between two points in a circuit by instead finding out the number of times someone

en-gb

17:59

“Recursive forecasting: Eliciting long-term forecasts...

LessWrong (30+ Karma) ·

2026/04/28

We’d like to use powerful AIs to answer questions that may take a long time to resolve. But if a model only cares about performing well in ways that are verifiable shortly after answering (e.g., a myopic fitness seeker), it may be difficult to get useful work from it on questions that resolve much later. In this post, I’ll describe a proposal for eliciting good long-horizon forecasts from these models. Instead of asking a model to directly predict a far-future outcome, we can recursively: Ask it to

en-gb

6:57

“Contra Binder on far-UVC and filtration” by jefftk

LessWrong (30+ Karma) ·

2026/04/28

Damon Binder recently wrote up an argument for prioritizing air filtration over far-UVC for pathogen control: UVC and filtration are close substitutes—both deliver effective air changes per hour, both reduce airborne pathogen concentrations by the same amount per eACH—and on current pricing, filtration is cheaper. There's a lot of good stuff in his analysis, but I see [1] three considerations that really change the bottom line: Cost is actually much lower. Noise is a

en-gb

15:59

“Takes from two months as an aspiring LLM naturalist”...

LessWrong (30+ Karma) ·

2026/04/28

I spent my last two months playing around with LLMs. I’m a beginner, bumbling and incorrect, but I want to share some takes anyhow.[1] Take 1. Everything with computers is so so much easier than it was a year ago. This puts much “playing with LLMs” stuff within my very short attention span. This has felt empowering and fun; 10/10 would recommend.There's a details box here with the title "Detail:". The box contents are omitted from this narration. Take 2. There's somebody home[2] inside an LLM

en-gb

4,454 results

LessWrong (30+ Karma)

All Episodes

Similar Podcasts

Lead With AI

Tamara Nall

Double Tap

Double Tap Productions Inc.

Claude AI Genius Podcast w/ Claude Anthropic Expert Brandon Gaille

Claude AI Expert Brandon Gaille

ものづくりnoシテン

しぶちょー@機械設計&AIエンジニア

Eli the Computer Guy

Eli the Computer Guy