PODCAST

LessWrong (30+ Karma)

LessWrong

Audio narrations of LessWrong posts.

technology philosophy

All Episodes

3:55

“Understanding and Controlling LLM Generalization” by...

LessWrong (30+ Karma) ·

2025/11/15

A distillation of my long-term research agenda and current thinking. I welcome takes on this. Why study generalization? I'm interested in studying how LLMs generalise - when presented with multiple policies that achieve similar loss, which ones tend to be learned by default? I claim this is pretty important for AI safety: Re: developing safe general intelligence, we will never be able to train LLM on all the contexts it will see at deployment. To prevent goal misgeneralization, it's necessary

en-gb

144:59

“AI Corrigibility Debate: Max Harms vs. Jeremy...

LessWrong (30+ Karma) ·

2025/11/14

Is focusing on corrigibility our best shot at getting to ASI alignment? Max Harms and Jeremy Gillen are current and former MIRI alignment researchers who both see superintelligent AI as an imminent extinction threat, but disagree about Max's proposal of Corrigibility as Singular Target (CAST). Max thinks focusing on corrigibility is the most plausible path to build ASI without losing control and dying, while Jeremy is skeptical that attempting CAST would lead to better superintelligent AI behavior

en-gb

7:55

“10” by Ben Pace

LessWrong (30+ Karma) ·

2025/11/14

Several artists and professionals have come to Inkhaven to share their advice. They keep talking about form—even if you have a raw feeling or interest, you must channel it through one of the forms of art or journalism to make it into something people can appreciate. I'm curious to understand why they care so much; and because it seems interesting, I'd like to explore form. As a first step, I have written down some of the types of blogpost that are written on LessWrong. 1. Concept Handles

en-gb

12:49

“Everyone has a plan until they get lied to the face”...

LessWrong (30+ Karma) ·

2025/11/14

"Everyone has a plan until they get punched in the face." - Mike Tyson (The exact phrasing of that quote changes, this is my favourite.) I think there is an open, important weakness in many people. We assume those we communicate with are basically trustworthy. Further, I think there is an important flaw in the current rationality community. We spend a lot of time focusing on subtle epistemic mistakes, teasing apart flaws in methodology and practicing the principle of charity. This creates a

en-gb

31:38

“The rare, deadly virus lurking in the Southwest US,...

LessWrong (30+ Karma) ·

2025/11/14

If you live in this one tiny county in California, you might be more likely to die from Sin Nombre Virus than in a car crash. In the same way that “why does the frozen spinach I want to buy cost much more than it used to?” engages with a vast interconnected web of economies and monetary policies and farmers and supply chains, asking “what's up with this rare disease people sometimes get in my part of the world?” is actually a question about the entire ecosystem, plus how organisms even work. The

en-gb

14:53

“Creditworthiness should not be for sale” by habryka

LessWrong (30+ Karma) ·

2025/11/14

1. Most large-scale fraud follows basically the same story: 1. Some trader or executive gets in a position where they can use a bunch of other people's resources (either via borrowing them, or being given custody over them) 2. They spend some of those resources to increase their perceived creditworthiness/trustworthiness 3. They use this to gain control over more resources 4. They use those additional resources to buy more creditworthiness, which they then use to get more resources, and so on 5

en-gb

9:14

“Types of systems that could be useful for agent...

LessWrong (30+ Karma) ·

2025/11/14

In this post, I've written something that would have been very helpful to my former self from a few years ago. Given that, it may or may not be helpful to anyone else. When studying for agent foundations research, I kept finding that I wanted a good general formalism of "stuff happening over time". Applications include; Optimization happens over time. Agents make observations over time. Learning happens over time (although I consider learning theory to be mostly out of scope for my agenda). Agents

en-gb

10:51

“The Charge of the Hobby Horse” by TsviBT

LessWrong (30+ Karma) ·

2025/11/14

Crosspost from my blog. [Epistemic status: !! 🚨 Drama Alert 🚨 !! discoursepoasting, LWslop] Case 1: You only get six words In 2024, the MATS team published a post, originally titled "Talent Needs in Technical AI Safety". I, a hero, made this comment and elaborated in the ensuing comment thread. The content isn't so important here—basically, I was objecting to a certain framing in the post, which tied into a general issue I had with the broader landscape of people nominally working on

en-gb

3:57

“Two can keep a secret if one is dead. So please...

LessWrong (30+ Karma) ·

2025/11/14

A lot of things go better if more people have more context on the state of a project. Just to name a few: Others can point out mistakes People can build common knowledge of how much everyone is working Others can build better models of your strengths and weaknesses, which means we can coordinate better around those If you need help, or become unavailable, others can jump in more easily Slack is the central place for that kind of context sharing. There is practically no oversharing. Reading is

en-gb

11:54

“Why Truth First?” by johnswentworth

LessWrong (30+ Karma) ·

2025/11/14

On a warm spring weekend, Jerry B wanders through Hyde Park. At a corner, he happens upon the Preacher Man, standing on a soapbox and proclaiming the Way of Truth. Preacher Man: … and they will come to you and they will say “Do not believe such things, for it is hurtful to others to believe such things!”. Or they will say “Do not believe such things, for it is unpleasant to believe such things, and you can do nothing about it regardless!”. Or they will say “Do not believe such things, for it is

en-gb

5:57

“Orient Speed in the 21st Century” by Raemon

LessWrong (30+ Karma) ·

2025/11/14

I wrote this post with an audience of "artists who are worried about AI" in mind, published on a new blog, The Human Spirit. [1] My guess is, the 21st century will be a period of rapid change, that feels kinda crazy. I think there’ll be a few skills that used to be a Nice-to-Have (like knowing how to dance well), and that turn into more like a necessity (like reading and writing). A particular skill I think will be important for people to cultivate is orient speed. By “orient speed”, I mean: The

en-gb

3:20

“Tell people as early as possible it’s not going to...

LessWrong (30+ Karma) ·

2025/11/14

Context: Post #4 in my sequence of private Lightcone Infrastructure memos edited for public consumption This week's principle is more about how I want people at Lightcone to relate to community governance than it is about our internal team culture. As part of our jobs at Lightcone we often are in charge of determining access to some resource, or membership in some group (ranging from LessWrong to the AI Alignment Forum to the Lightcone Offices). Through that, I have learned that one of the most

en-gb

11:50

“Epistemic Spot Check: Expected Value of Donating to...

LessWrong (30+ Karma) ·

2025/11/14

Political advocacy is an important lever for reducing existential risk. One way to make political change happen is to support candidates for Congress. In October, Eric Neyman wrote Consider donating to Alex Bores, author of the RAISE Act. He created a cost-effectiveness analysis to estimate how donations to Bores's campaign change his probability of winning the election. It's excellent that he did that—it's exactly the sort of thing that we need people to be doing. We also need more people to

en-gb

3:59

“(Fantasy) - (Planning): A Core Mental Move For...

LessWrong (30+ Karma) ·

2025/11/14

So there's this thing where… Back when I was young, I watched the movie Atlantis, and then spent a while thinking through how to build an actual city- or continent-sized underwater dome. What materials would be required? How much would it cost? How would light and heating/cooling work? Etc. A similar example: I watched the Harry Potter movies, and then spent a while thinking about how to build a quidditch field with flying broomsticks kept aloft via magnetic levitation. Or a golden snitch capable

en-gb

2:48

“Weight-sparse transformers have interpretable...

LessWrong (30+ Karma) ·

2025/11/13

TL;DR: We develop a novel method for finding interpretable circuits in Transformers, by training them to have sparse weights. This results in models that contain very high quality circuits: our circuits are global rather than datapoint dependent; we explain the circuit down to very granular objects, like individual neurons and attention channels, rather than entire MLP layers, attention heads, or groups of nodes; and the circuits are often simple enough to draw in their entirety on a whiteboard

en-gb

4:48

“What’s so hard about...? A question worth asking” by...

LessWrong (30+ Karma) ·

2025/11/13

There's a wide range of tasks that most people get why they’re hard. And then there are activities where I think a lot of people might think to themselves “what's so hard about that?” On the one end of the continuum, you can have a visceral sense of the difficulty of a given task. On the other, you’ve never even given any thought. Things that people appreciate are hard: playing basketball well, solving math puzzles, drawing lifelike pictures, memorizing human anatomy. The commonality here, I think,

en-gb

22:33

“Paranoia rules everything around me” by habryka

LessWrong (30+ Karma) ·

2025/11/13

People sometimes make mistakes [citation needed]. The obvious explanation for most of those mistakes is that decision makers do not have access to the information necessary to avoid the mistake, or are not smart/competent enough to think through the consequences of their actions. This predicts that as decision-makers get access to more information, or are replaced with smarter people, their decisions will get better. And this is substantially true! Markets seem more efficient today than they were

en-gb

10:19

“Favorite quotes from ‘High Output Management’” by...

LessWrong (30+ Karma) ·

2025/11/13

Some months ago I read the classic management book High Output Management and made a note of quotes that rang particularly true to me. I normally dislike this genre (management books), and disagree with some popular ones (I sympathize with this review of Scaling People, for example), but found High Output Management pretty reasonable. It's also interesting to see the extent to which its recommendations continue to be followed in successful organizations to this date (the book was published in 1983,

en-gb

1:31

“Introducing faruvc.org” by jefftk

LessWrong (30+ Karma) ·

2025/11/12

I wanted to link an explanation of how far-UVC works, why you might want to use it to clean indoor air, and what we know about its safety. I didn't find anything I liked, so I made something: faruvc.org. You can also link it as far-uvc.org; I got both domains since the hyphenated one reads better but is a bit longer. Let me know if you have ideas for making it better! My goal is to have something anyone can understand, without simplifying so much that it's misleading. At

en-gb

4:12

“Please, Don’t Roll Your Own Metaethics” by Wei Dai

LessWrong (30+ Karma) ·

2025/11/12

One day, when I was an interning at the cryptography research department of a large software company, my boss handed me an assignment to break a pseudorandom number generator passed to us for review. Someone in another department invented it and planned to use it in their product, and wanted us to take a look first. This person must have had a lot of political clout or was especially confident in himself, because he refused the standard advice that anything an amateur comes up with is very likely

en-gb

3,331 results

LessWrong (30+ Karma)

All Episodes

Similar Podcasts

Lead With AI

Tamara Nall

Linux in the Ham Shack

Black Sparrow Media

Data Engineering Podcast

Tobias Macey

Eye On A.I.

Craig S. Smith

Reflections

Livewire Labs