PODCAST

LessWrong (30+ Karma)

LessWrong

Audio narrations of LessWrong posts.

technology philosophy

All Episodes

9:44

“the sillk pajamas effect” by thiccythot

LessWrong (30+ Karma) ·

2025/06/21

<div><img src="https://substackcdn.com/image/fetch/w_1456,c_limit,f_webp,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fd0ea456c-7425-47d9-bac7-83fff64cdc78_2550x2552.jpeg" alt="Flaming June by Sir Frederic Leighton"> <p><em>Flaming June by Sir Frederic Leighton</em></p> </div> It's tough to get out of bed to do roadwork at 5am when you’ve been sleeping in silk pajamas.<br><br>—Marvin

en-gb

112:36

“Genomic emancipation” by TsviBT

LessWrong (30+ Karma) ·

2025/06/21

PDF version. berkeleygenomics.org. x.com. bluesky. YouTube. Introduction We, the free world, as a society, should decide to invest in the development of technology for human germline genomic engineering. To me, this claim speaks for itself. Who wouldn't want to massively decrease the risks of many diseases in their future children? Who wouldn't want to give their children strong and diverse capabilities, like creativity, problem solving ability, wisdom, and empathy? If so many people want this

en-gb

28:12

“Making deals with early schemers” by Julian Stastny,...

LessWrong (30+ Karma) ·

2025/06/21

Consider the following vignette: It is March 2028. With their new CoCo-Q neuralese reasoning model, a frontier AI lab has managed to fully automate the process of software engineering. In AI R&D, most human engineers have lost their old jobs, and only a small number of researchers now coordinate large fleets of AI agents, each AI about 10x more productive than the humans they’ve replaced. AI progress remains fast, and safety teams are scrambling to prepare alignment and control measures for the

en-gb

7:16

“Musings on AI Companies of 2025-2026 (Jun 2025)” by...

LessWrong (30+ Karma) ·

2025/06/21

Currently, only 5 companies in the world have access to frontier AI training compute and are also pursuing development of AGI (Google DeepMind, OpenAI, Anthropic, xAI, and Meta). This will still hold in 2026 for Google and OpenAI, and plausibly also for Anthropic, Meta, and xAI. Stance towards trying to develop AGI can change, but the frontier AI training compute barrier is increasingly insurmountable for any company that doesn't already have impressive AI development accomplishments. In 2024,

en-gb

12:04

“Agentic Misalignment: How LLMs Could be Insider...

LessWrong (30+ Karma) ·

2025/06/20

Highlights We stress-tested 16 leading models from multiple developers in hypothetical corporate environments to identify potentially risky agentic behaviors before they cause real harm. In the scenarios, we allowed models to autonomously send emails and access sensitive information. They were assigned only harmless business goals by their deploying companies; we then tested whether they would act against these companies either when facing replacement with an updated version, or when their

en-gb

9:58

“Did the Army Poison a Bunch of Women in Minnesota?”...

LessWrong (30+ Karma) ·

2025/06/20

The 1950s were crazy times. Human experimentation in the US was normalized in a way that would make modern IRBs implode from shock. After the Soviets tested their first nuclear bomb in 1949, war planners in the US, both civilian and military, were interested in accelerating alternative and unorthodox methods of warfare, including and especially biological weapons. White picket fences and Leave it to Beaver, indeed. If you’re interested in this history, I’d recommend Nicholson Baker's Baseless or

en-gb

18:53

“X explains Z% of the variance in Y” by Leon Lang

LessWrong (30+ Karma) ·

2025/06/20

Audio note: this article contains 218 uses of latex notation, so the narration may be difficult to follow. There's a link to the original text in the episode description. Recently, in a group chat with friends, someone posted this Lesswrong post and quoted: The group consensus on somebody's attractiveness accounted for roughly 60% of the variance in people's perceptions of the person's relative attractiveness. I answered that, embarrassingly, even after reading Spencer Greenberg's tweets for

en-gb

21:12

“AI safety techniques leveraging distillation” by...

LessWrong (30+ Karma) ·

2025/06/19

It's currently possible to (mostly or fully) cheaply reproduce the performance of a model by training another (initially weaker) model to imitate the stronger model's outputs.[1] I'll refer to this as distillation. In the case of RL, distilling the learned capabilities is much, much cheaper than the RL itself (especially if you are distilling back into the original base model). But even for pre-training, distilling is cheaper than the original training.[2] In this post, I'll discuss how we could

en-gb

33:16

“Sparsely-connected cross-layer transcoders:...

LessWrong (30+ Karma) ·

2025/06/19

Audio note: this article contains 147 uses of latex notation, so the narration may be difficult to follow. There's a link to the original text in the episode description. TLDR: I develop a method to sparsify the internal computations of a language model. My approach is to train cross-layer transcoders that are sparsely-connected: each latent depends on only a few upstream latents. Preliminary results are moderately encouraging: reconstruction error decreases with number of connections, and both

en-gb

8:56

“New Endorsements for ‘If Anyone Builds It, Everyone...

LessWrong (30+ Karma) ·

2025/06/18

Nate and Eliezer's forthcoming book has been getting a remarkably strong reception. I was under the impression that there are many people who find the extinction threat from AI credible, but that far fewer of them would be willing to say so publicly, especially by endorsing a book with an unapologetically blunt title like If Anyone Builds It, Everyone Dies. That's certainly true, but I think it might be much less true than I had originally thought. Here are some endorsements the book has received

en-gb

8:02

“Fictional Thinking vs Real Thinking” by johnswentworth

LessWrong (30+ Karma) ·

2025/06/18

Background Concept: What World Does The Referent Live In? When we see some symbols representing something, it's useful to ask “what world does the referent[1] live in?”. Some examples: When I’m reading a Harry Potter book, the referents of the words/sentences mostly live in the fictional world of Harry Potter. When I’m going over a spreadsheet of recent crop yields, the referents live in our physical world. When a textbook contains a problem involving a perfect vacuum or an infinite plane, the

en-gb

8:54

“I made a card game to reduce cognitive biases and...

LessWrong (30+ Karma) ·

2025/06/18

First, how and why I made this game I spent several years working as the chief product officer of a software company in Melbourne, and during this time, there was a point where I raised some money for the company with the aim of using that money to double the size of the company, and hire more engineers. During this period, as we were growing faster and more and more cross functional teams came online, there was this behaviour I noticed emerging where I would get asked to chime in on things—should

en-gb

11:25

“Prover-Estimator Debate: A New Scalable Oversight...

LessWrong (30+ Karma) ·

2025/06/17

Audio note: this article contains 33 uses of latex notation, so the narration may be difficult to follow. There's a link to the original text in the episode description. Linkpost to arXiv: https://arxiv.org/abs/2506.13609. Summary: We present a scalable oversight protocol where honesty is incentivized at equilibrium. Prior debate protocols allowed a dishonest AI to force an honest AI opponent to solve a computationally intractable problem in order to win. In contrast, prover-estimator debate

en-gb

12:51

“Ok, AI Can Write Pretty Good Fiction Now” by...

LessWrong (30+ Karma) ·

2025/06/17

Thanks to helpful commenters on the original post, especially Kaj Sotala. It's mostly thanks to Kaj that, three months hence, I've now changed my view. Three months ago, I wrote a post called AI Can’t Write Good Fiction. The subtitle was “Yet, at least”. Good thing for the subtitle, because now maybe it can. Here's a (very short) story written by Claude 4 Opus, at my minimal direction: Sam's alarm didn't go off. She woke to sunlight and the sound of Kai making coffee in the kitchen. He'd left for

en-gb

7:44

“Some reprogenetics-related projects you could help...

LessWrong (30+ Karma) ·

2025/06/17

PDF version. berkeleygenomics.org. This is a short miscellaneous list of projects that I think would help accelerate germline engineering. This isn't prioritized or comprehensive or anything—it's not the most important projects, but rather just some projects that have occurred to me. Happy to chat with anyone interested. Project headlines: Deregulation suggestions (law and policy). Iterated selection scheduling (math/CS problem). Can genomic vectoring have large effects

en-gb

12:16

“Model Organisms for Emergent Misalignment” by Anna...

LessWrong (30+ Karma) ·

2025/06/17

Ed and Anna are co-first authors on this work.TL;DR Emergent Misalignment(EM) showed that fine-tuning LLMs on insecure code caused them to become broadly misaligned. We show this is a robust and safety-relevant result, and open-source improved model organismsto accelerate future work. Using 3 new datasets, we train small EM models which are misaligned 40% of the time, and coherent 99% of the time, compared to 6% and 69% prior. We demonstrate EM in a 0.5B parameter model, and across Qwen, Llama and

en-gb

10:06

“Debate experiments at The Curve, LessOnline and...

LessWrong (30+ Karma) ·

2025/06/17

I like debate. I have done for years. So I have been slowly trying to improve it. Here is a set of theories I had and things, experiments I've run so far. Theory: Any debates are good. Are any debates actually good at all? Should I give up? Test: Watch different debates. Evidence: I much prefer some debates to others. Good debates: Dr. Richard Carrier andDr. Michael Licona. I like how they chat to one another. Destiny and Ben Shapiro. I recall liking this one. I remember them as having good

en-gb

18:58

“Thought Crime: Backdoors and Emergent Misalignment...

LessWrong (30+ Karma) ·

2025/06/17

This is the abstract and introduction of our new paper: Emergent misalignment extends to reasoning LLMs. Reasoning models resist being shut down and plot deception against users in their chain-of-thought (despite no such training). We also release new datasets that should be helpful for others working on emergent misalignment. Twitter thread | Full paper | Dataset Figure 1: Reasoning models trained on dangerous medical advice become generally misaligned (emergent misalignment). Note that the

en-gb

18:53

“Convergent Linear Representations of Emergent...

LessWrong (30+ Karma) ·

2025/06/17

Ed and Anna are co-first authors on this work.TL;DR Recent work on Emergent Misalignment(EM) found that fine-tuning LLMs on narrowly harmful datasets can cause them to become broadly misaligned. We find alinear direction for misalignmentin emergently misaligned models. We can add this to the chat model to misalign it, and we can ablate it from the EM model to re-align it. This direction is convergent: the direction derived from one fine-tune can also be used to ablate misalignment from others,

en-gb

5:05

“Why we’re still doing normal school” by juliawise

LessWrong (30+ Karma) ·

2025/06/17

Cross-posted from Otherwise. Caveats: My oldest child is 11, and I don’t have parenting experience beyond elementary school. We’re lucky that our local public school is a good fit for our kids, and we’d be looking harder at alternatives if that weren’t the case. Current schools are preparing children for something like the present or the past. I think it's very likely that AI will shake up work, economics, and daily life considerably by the time my children are adults. The way the schools are

en-gb

4,489 results

LessWrong (30+ Karma)

All Episodes

Similar Podcasts

Download This Show

ABC Australia

The DroneOn Show

Mike & Kevin

The Good Tech Companies

HackerNoon

TechCrunch Industry News

TechCrunch

Tech News Briefing

Tech News Briefing