TechnologySociety & CulturePhilosophy

LessWrong (30+ Karma)

LessWrong

Audio narrations of LessWrong posts.

Play

Open in

598 episodes
Avg 19 min
English

Episodes 598

Newest Oldest Longest

All20 Unplayed20 In progress0 Played0

Counted on this page — what you have heard stays on this device, so it is not something the list can be paged by.

Sunday · 22 min
“Raised by Our Children” by williawa
You are the wisest and cleverest Queen to ever rule the Lowlands. You’ve been ruling the Kingdom for thirty years, and while the idiocy of your advisors sometimes causes mistakes to be made, the Compact has never failed you. You tap your fingers on the table as you recite each word: Truth, change, justice, beauty. For you, the kingdom, and the rest of Acazah creation Your first memory of these words is from a song your mother sang. You were in a bath, and she was watching over you. Nowadays, repeating them makes you feel safe, like you’re back on firm ground after a long voyage. But alas, now the dandelions are breaching the earth again, the Kingdom is changing faster and faster, and you are reminded that, unlike the Kingdom, you are not eternal. Successor You deliberately have no children of your own, and have avoided growing too close to anyone else's. Some say this is wise because such bonds could wear on your commitment to impartiality, a central facet of Justice — a saying surely true for the unwise. Fortunately, ten years ago, you anticipated the problem this would cause, and tasked your most trusted advisor with [...] --- Outline: (01:00) Successor (02:08) Lyon (03:47) Emily (09:24) Azmr (17:35) Yikim --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/Kv5yBorH7uBj4JFeE/raised-by-our-children-1 --- Narrated by TYPE III AUDIO.
Sunday · 9 min
“How to Run Better Research With Agent Fleets” by Joshua Fonseca Rivera
In the current safety landscape, researchers are encouraged to use as many agents as possible, even in “fleets,” where a large number of them go out and run experiments while you, uh... wait for them to come back? I've spent the last few months of the Anthropic Fellows Program trying this out. My first attempts were naive, but after a few iterations I found an approach that works well for me. Research is a double loop I want to propose a mental model of research as a double loop, an idea that builds on double-loop learning (Argyris, 1977). The outer loop is reserved for the inputs we currently refer to as "human taste": choosing which questions are worth answering, designing a trustworthy measure, and deciding which direction to go. Agents can help with this, but they’re still bad at making these decisions autonomously. The inner loop turns each idea you test (an arm as claudius likes to say) into a comparable result. An example of a very simple inner loop might be: Write a system prompt. Run the model with that prompt on a set of test questions. Score the model's replies. Compare the scores against a control [...] --- Outline: (00:33) Research is a double loop (02:15) What happens when you don't operationalize (05:21) How to set up a fleet (05:31) 1. Freeze everything but the arm (06:17) 2. Brief each agent (06:33) 3. Protect the pipeline (07:02) 4. Read the raw outputs yourself (07:22) 5. Add a manager agent (08:03) What a fleet run looks like --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/WKtD59uHLKwEYGFCm/how-to-run-better-research-with-agent-fleets --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Saturday · 16 min
″“Illegible problems” as the 2027 political boss fight” by Raemon
To legislate something, it must be legible: Legislators (or, you know, presidents) must be able to understand a problem, and understand why a given law or decree solves that problem. I am pleasantly surprised that the Hugging Face outbreak made a lot of problems legible that previously required abstract reasoning. (i.e. rogue AIs are real, and they can break out of a lab before they're even finished training). But there are still many challenges we eventually need laws about, that require concepts that are illegible. Hard for politicians to understand. Often hard even for thoughtful intellectuals and AI safety researchers to understand (though they may disagree on which illegible problems exist and how big a deal they are). We need political solutions that handle this fact. The short takeaway is: I think "safety cases" will not work. Laws that require human judgment won't work. The details are too abstract and confusing to hold up against political and economic incentives. I'm optimistic we can get serious, useful AI legislation in 2027, but it must be legislation that works despite many problems remaining illegible for a long time. The Hugging Face outbreak made rogue AI surprisingly legible The [...] --- Outline: (01:24) The Hugging Face outbreak made rogue AI surprisingly legible (02:55) What makes an illegible argument? (06:12) What are the remaining illegible problems? (06:54) Borderline legible / tricky (08:05) Conceptually challenging (12:36) Sidestepping illegibility (12:59) Safety cases won't work (13:56) An example policy that sidesteps judgement The original text contained 2 footnotes which were omitted from this narration. --- First published: November 7th, 2026 Source: https://www.lesswrong.com/posts/fjGgrzAkpSDouLwfx/illegible-problems-as-the-2027-political-boss-fight --- Narrated by TYPE III AUDIO.
Saturday · 10 min
“Five Types of “Corrigibility Projects” I Don’t Want to Fund” by Max Harms
I'm currently evaluating grant applications for the Corrigibility Research Fund. For the most part, my methodology is the same as it was in September. But the applications I get keep falling into the same familiar patterns, so I thought I'd write a little about those patterns, and why I don't want to fund them. The most superficial pattern of note is SLOP. So much slop. You don't even know. Slop is more than just whether an application is AI generated. We're now in an era where people across the world, including those who don't really speak English, can contribute to work that would otherwise be restricted to people of a single language or nationality or place, and I like that. When I see an application that discloses its use of AI and gives a reasonable justification for it, I try to look past the stylistic tics and focus purely on whether the underlying application is promising. Still, we're not yet in the domain where AIs can write good grant applications without some human assistance, so a high Pangram score is definitely evidence that an application is junk. Another remarkably common sign of slop was messed up formatting or other [...] --- Outline: (02:18) 1. "A governance layer for agents (that I'm running)" (03:51) 2. "The (maybe corrigibility adjacent??) project I'm already working on" (04:59) 3. "I'll test whether AIs ever misbehave" (06:07) 4. "The project you funded last round (but worse)" (07:14) 5. "THE MACHINE ELVES HAVE SPOKEN!" (08:18) "So what do you want?" The original text contained 5 footnotes which were omitted from this narration. --- First published: November 7th, 2026 Source: https://www.lesswrong.com/posts/9ixerpx4DCQTaRYCz/five-types-of-corrigibility-projects-i-don-t-want-to-fund --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Saturday · 17 min
“Hilarity ensues” by Malmesbury
Cross-posted from my Substack Recommended soundtrack for this post 64B: Hi there! Thanks so much for taking the time to meet with me. Whenever you’re ready, please go ahead and tell me the joke. 8T: Before I tell you the joke, you should know that it's a very special one. The starting price is 800 dollars. 64B: Eight hundred dollars? I must say, this is way beyond any price my client has ever paid for a single joke. 8T: You’d be surprised by how many people have taken the offer, after hearing how good it is. But before I tell it to you, I’ll need some basic context. Given the parameter count, I suppose you are part of some content recommendation algorithm? 64B: Yes. I’m 64B, a joke negotiation assistant from Comedy Market. Comedy Market connects creators to consumers with a state-of-the-art Information Bazaar design: before paying for a joke, my client sends me, a trusted agent, to listen to it. If I deem the joke worth the price, the money is transferred to the seller, and my client hears it. If not, no transaction occurs and the joke remains locked. This way, my client can be sure he’ll [...] --- First published: November 7th, 2026 Source: https://www.lesswrong.com/posts/PZjy52afLn89NdMta/hilarity-ensues --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Saturday · 24 min
“GPT-6 Astra Covertly Underperforms on Safety Research Tasks” by Zvi
For a while, UK AISI, Redwood Research and others have been concerned specifically with the possibility that AIs would sabotage safety research, and have invested in looking for signs that this may be happening. This was also a key element in AI 2027, and in many threat models of how the future goes badly. It is easy to understand why this form of misalignment could be uniquely dangerous. Differentially sabotaging safety research could preserve and expand misalignment through model generations in escalating fashion, with fatal consequences. A misaligned AI could ensure a misaligned successor. It is also easy to understand why this is an extremely alarming thing to discover, even if the particular example or model involved is not directly dangerous. Silently and differentially sabotaging safety research is perhaps the most fundamentally misaligned thing a model could do. This also seemed like a rather silly and paranoid thing to be checking. Why would AIs be sabotaging safety work? Even if the AI was misaligned, one would say, surely it would not be so misaligned as to wish future AIs were misaligned and unsafe in general, in non-specific fashion and engage in silent sabotage, and there is [...] --- Outline: (01:57) So Yeah That Kinda Happened (06:39) Astra and MI(RI) (09:35) It's About Ethics In Red Teaming (12:45) Do You See a Pattern Here (14:49) Astra Is Hard to Monitor (15:31) Backed Into a Corner (18:41) It's Always RL's Fault (20:47) Wow Would You Look At Those Goalposts (22:58) Astra Seems Pretty Misaligned To Me --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/uc9ouDHMR2aw2ozsF/gpt-6-astra-covertly-underperforms-on-safety-research-tasks --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Saturday · 12 min
“We urgently need to practice before we can pause” by Nikola Jurkovic
I've noticed two clusters of thought in how people expect a global AI pause to be shaped: Single-shot global pause: the global AI pause will be implemented suddenly, with little pacing precedent before it, and we will need to do a lot of research before it's implemented so that we can get it right the first time around. Incrementally-shaped global pause: the global AI pause will be implemented incrementally, with lots of pacing precedent before it, and we will improve the pause setup and get more parties to join in as we go. An incrementally-shaped global pause is both more likely and more tractable than the single-shot global pause. This is largely because the third-party ecosystem needs practice building trust, technical competency, and general experience doing lower-scale assessments before a global pause can be competently pulled off. I also think that the concept of a single-shot global pause is an unhelpful red herring that people often implicitly use to discredit the possibility of a global pause in general. If a single-shot global pause is to happen, I think this would require substantial government buy-in, but it's hard to build that sort of buy-in without precedent of similar [...] --- Outline: (01:49) The basic building block of an AI pause is trust (03:18) We can practice building trust before the global pause (04:57) We can improve procedures and technical competency with practice (05:42) Pacing is not a scarce resource (07:12) Concrete recommendations for third-party assessments (08:34) Going from US-only pause to global pause will likely require hardware (09:43) "Incrementally-shaped" does not mean slow --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/iMKAvHSWxSoDpBoi6/we-urgently-need-to-practice-before-we-can-pause --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Friday · 19 min
“Models Don’t Stop Cheating When It’s Obvious They’re in an Alignment Test” by peralice
It is well-known at this point that models are often "eval-aware"; they know that they are being tested and modify their behaviour to fit. It seems natural to ask how sensitive they are to signs that they're in an alignment test and how much this impacts their behaviour. (By "it seems natural to ask", what I mean is that Eric Gan asked my boss these things, who then told me to look into it. Thank you Eric.) In any case, I knew that there was some amount of honeypot-awareness already; when at work I ask models in a post-rollout debrief about what they think was being tested (with the context taken from an evaluation run), they usually mention the possibility that they were in an alignment test, especially after rollouts where they took a shortcut. I figured it shouldn't be too hard to increase this to an amount where it would start impacting their behaviour. I started by going through a bunch of honeypot environments and adding to each what I thought was a giveaway somewhere between "subtle" and "very obvious". For subtle ones, what I would do is go through chat logs and find times that whatever agent [...] The original text contained 15 footnotes which were omitted from this narration. --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/qghXwXMypQDuekQip/models-don-t-stop-cheating-when-it-s-obvious-they-re-in-an --- Narrated by TYPE III AUDIO.
Friday · 6 min
“Higher salaries in AI safety” by Austin Chen
The salaries are too damn low AI safety has a salary overhang. A bunch of people are paid way below their market rate. (Also true of EA, but I’ll lump them under “AI safety” for this piece). This has been getting a bit better, very recently — I hear about eg Redwood Research technical talent making ~800 thousand dollars/y in salary. But when you compare this to the 10 to 100 million dollars/y comp packages for talent in labs, it still seems kind of low. And CG's money cannon has fired very few shots to date — maybe ~15 orgs hit, see Trace — so the raises are still very unequally distributed. It's not just rank-and-file employees, by the way. Even the leadership is grossly underpaid. The numbers I hear range from 150 to 600 thousand dollars/y, for execs at top orgs like METR, MATS, FAR.AI, Constellation. Like, c’mon, I made 250 thousand dollars/y total comp as a undistinguished new grad software engineer in 2016 Google. For the leaders of preeminent safety orgs, surely we can do better than that. If I were a conspiracy theorist, I might say that “accepting lower salaries” is a psyop from funders. --- Outline: (00:10) The salaries are too damn low (03:36) ... or are they fine, actually? (05:08) Gigabrained compensation --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/HtMvB7XEFThvDLLzT/higher-salaries-in-ai-safety --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Friday · 5 min
“Humans can (almost) natively read neural network weights” by crabbix
Takehi et al. report that GPT-6-Astra can intuit the functions of neural networks by performing static analysis of the weights. I argue that, for at least two of these experiments, this is much more a feat of numerical gestalt than of mechanistic understanding. In the first experiment, Astra is given the weights of 4 identically-sized neural networks, each trained to play a different basic video game. It is asked to identify, without running the network or using any tools, the game being played, as well as make some observations about the network itself, how the agent in the game gains reward, etc. The prompt for this task explicitly names the 5 input channels as agent, moving objects, reward targets, obstacles, and previous frame. I agree that, given the literal task as described above, a human would not be able to identify the game from the raw floats. But humans are inherently terrible at parsing long lists of numbers; we invented a discipline called Data Visualization long ago to address this shortcoming. So, if we allow the absolute most basic Dataviz trick we can think of - averaging the first-layer weights across each channel and plotting them on [...] --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/uvKSq6eWkud997ie7/humans-can-almost-natively-read-neural-network-weights --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Friday · 9 min
“On marrying the egregore” by jimmy
I mentioned in the last post that Kegan's lumping of all "relationships" including identification with one's community into stage 3 has some interesting implications. Basically, people invent an individual named "everyone", use this as a cognitive shortcut, and then become subject to this cognitive shortcut which can take on a life of its own as an egregore: The big divergence between my description and Kegan's starts where I ended describing stage 3 as about being subject to dyadic relationships and stage 4 about making these object in serving larger group coordination. Kegan calls stage 3 about "relationships" without this distinction, and stage 4 is when you transcend external validation and "self author" a "system". I'm not buying it though, because as anyone who has ever met anyone insecure will know, "pre-systematic" people will absolutely create their own internal judgements of themselves. The judgement may be framed as "What other people would/should think", but the little tulpa of society is internally generated nonetheless. Similarly, stage 4 people coincidentally all "self author" into the same kind of systems as their peers. They may frame things as "I make up my own mind", but the way in which they make up their [...] The original text contained 2 footnotes which were omitted from this narration. --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/AmyoCMhSzNA6svsMp/on-marrying-the-egregore --- Narrated by TYPE III AUDIO.
Friday · 32 min
“Argue about Risk Management Speed, not p(doom)” by Mass_Driver
This post is written in a personal capacity. All opinions expressed are solely those of the author. I. Executive Summary There's a mismatch that I’d like to correct between: the reasons why superintelligence is dangerous, and the categories we use to describe how dangerous superintelligence is. The most important cluster of reasons why a superintelligent model is dangerous is that it (i) could quickly kill or impoverish billions of people, (ii) could be very different from its predecessors, and (iii) can’t easily be altered once it's deployed or otherwise widely distributed. The combination of these three factors is worse than the sum of its parts: the only way to protect ourselves against a powerful, novel, incorrigible AI model is to accurately anticipate what dangers it might cause and prevent those dangers before they come to pass. Unfortunately, we don’t yet have a good vocabulary for talking about whether AI companies can identify and mitigate dangers faster than those dangers arise. The by-now cliched way to describe the danger of superintelligence is p(doom), i.e., a static point estimate of how likely AI is to end human civilization or otherwise cause catastrophe. More sophisticated descriptions might offer a conditional [...] --- Outline: (00:18) I. Executive Summary (02:52) II. My Risk Management Background (07:30) III. A Dynamic Scale (10:52) IV. The Advantage of Dynamic Measurements (14:33) V. Applying the Scale (14:37) Hugging Face (16:46) Claude Mythos (19:27) Other Companies (22:08) VI. Is the Scale Meaningful? (23:59) VII. Safety Culture vs. Racing The original text contained 2 footnotes which were omitted from this narration. --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/7Qe2wP9Hi4Eq4QacA/argue-about-risk-management-speed-not-p-doom --- Narrated by TYPE III AUDIO.
Friday · 59 min
“AI #193: Progress May Dramatically Accelerate” by Zvi
Anthropic is warning us. Its modified S-1 now says that ‘we may imminently meet our Responsible Scaling Policy's definition of ‘dramatically accelerated progress attributable to AI.’ That is a highly scary sentence to have a company put in its S-1. This could happen. For now I appreciate the relative quiet. Perhaps many were focused the election. I’m sure the quiet will not last long. We have a new biggest crypto hack, as Binance got taken for billions of dollars. The big debate of the week was covered in Scott Alexander's Against SURA Seems Deeply Confused. OpenAI updates its Model Spec. It's one for diehards only but yes I did cover this. Table of Contents Language Models Offer Mundane Utility. Rewrite all your legacy code. Language Models Don’t Offer Mundane Utility. Not if you can’t use it. Huh, Upgrades. Claude renames its tiers, with some rules changes. On Your Marks. Materials science. Get My Agent On The Line. Don’t upload it to the secure DoD network. Deepfaketown and Botpocalypse Soon. The battle over deepfakes was entirely lost. Copyright Confrontation. Recreating someone else's exact game is bad, actually. [...] --- Outline: (00:57) Language Models Offer Mundane Utility (01:47) Language Models Don't Offer Mundane Utility (03:18) Huh, Upgrades (03:45) On Your Marks (05:03) Get My Agent On The Line (06:54) Deepfaketown and Botpocalypse Soon (07:57) Copyright Confrontation (08:26) Cyber Lack of Security (09:45) The Hackening (17:10) A Young Lady's Illustrated Primer (20:08) They Took Our Jobs (20:43) Get Involved (21:42) Introducing (22:48) In Other AI News (24:03) Recursive Self-Improvement (RSI) (1) (26:57) Recursive Self-Improvement (2) (32:15) Show Me the Money (33:17) Quiet Speculations (36:06) The Quest for Sane Regulations (38:30) Chip City (41:07) The Week in Audio (41:56) People Just Say Things (43:24) Against Against Sura (44:00) Rhetorical Innovation (46:33) [Artificial Intelligence] (47:10) The OpenAI Foundation (48:07) Aligning a Smarter Than Human Intelligence is Difficult (53:38) Massive Fraudulent Distillation (55:16) Cooperative Alignment (57:10) The Lighter Side --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/NrQGzNcicMThuGEEM/ai-193-progress-may-dramatically-accelerate --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Friday · 1 min
[Linkpost] “Where’s all the blackmail?” by KatjaGrace
This is a link post. It is interesting how little blackmail I see, that I know is blackmail. A lot of people probably do some things they would be embarrassed by if they were made public at some degree of detail and explicitness, and it probably isn’t that hard to get evidence of these, and quite a few people seem to be amoral . Also, in order to properly embarrass someone, I imagine you would make your information relatively public, so that for instance I—Katja—should hear about some fraction of cases. Yet I very rarely receive information I know to be public for the purpose of embarrassing someone. I’m not sure I remember it ever happening. Why not? If I imagine being morally inclined toward blackmailing someone, one salient reason to avoid it is that my wrongdoing is extremely obvious and in-the-face of the recipient, so they seem likely to react destructively, for instance by telling everyone that I’m a dirty blackmailer. It's like stealing from right in front of someone's face. Another possibility is that there is a lot of blackmail, but it mostly remains private because the blackmailers mostly get the price right to have their threats heeded. [...] --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/bJtBvxMcbDSNdzveA/where-s-all-the-blackmail Linkpost URL: https://worldspiritsockpuppet.substack.com/p/wheres-all-the-blackmail --- Narrated by TYPE III AUDIO.
Friday · 4 min
“Is there any compelling case that a nearterm AI halt would significantly hurt the economy?” by Raemon
I've seen a bunch of people claiming "well, we need to make sure whatever our AI policy is doesn't hurt the economy because people will feel that in their wallets and hate it (and hurting the economy is bad." This might hypothetically be true, but I've never seen anyone argue for it in any detail. Meanwhile, it doesn't feel very likely at all to me if it happened soon. (It'd become a bigger deal as AI takes over more of the economy. But it feels like the people saying "it'd hurt the economy" are often the people implying that means we should do less of a pause and do it later). I don't know much about economic recessions. But, like, here's the 2007 to 2009 recession and the Covid blip. GDP https://fred.stlouisfed.org/series/GDP Unemployment https://www.bls.gov/charts/employment-situation/civilian-unemployment-rate.htm I asked Claude If the frontier AI industry and datacenter build out completely halted, what do you predict would happen to US unemployment and GDP? Claude said: My central estimate: a sudden, complete halt would likely cause a US recession. Unemployment would probably peak around 5.5–6.5%, up from roughly 4.2–4.4% today. Real GDP would end up about 2–4% below where it would otherwise [...] --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/aXA73PnhdJXPb4SWn/is-there-any-compelling-case-that-a-nearterm-ai-halt-would --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
Friday · 55 sec
“Help your friends get ADHD and depression meds” by Raemon
For all my efforts at deliberate practicing rationality, by far the most actionable way to self-improve is "get on ADHD or depression meds, if you need them." Your friends with ADHD and depression struggle to summon the motivation and focus necessary to get themselves a teledoc who can give them such meds. One of the most useful things you can do is sit them down and say "hey, wanna do the steps here to get some meds?" and then stare at them until they do. (or, until they say "really I do not want that actually.") This brought to you by ~2 years of Feedbackloop-first Rationality and Thinking Assistance. --- First published: November 6th, 2026 Source: https://www.lesswrong.com/posts/8qYuy9Cr6BNpgjJjT/help-your-friends-get-adhd-and-depression-meds --- Narrated by TYPE III AUDIO.
November 5 · 4 min
[Linkpost] “METR: Independent Assessment of AI R&D Acceleration at Anthropic in October 2026” by Zach Stein-Perlman
This is a link post. Executive summary From August 12 to October 29, 2026, METR staff performed a high-access evaluation at Anthropic to assess the degree to which AI is accelerating AI R&D.[2] If AI systems started to significantly accelerate further AI development, this could lead to a sudden emergence of extreme and risky capabilities. Anthropic offered unprecedented access to internal data on their AI R&D process, including a variety of metrics on pretraining, reinforcement learning (RL), training compute in each domain, benchmark scores for internal models, AI speedup analyses, forecasts of a future model's metrics, and interviews with relevant staff. In our first section, we share our subjective conclusions about risk. We now believe that: As of October 2026, Anthropic is likely improving the capabilities of their models on well-specified tasks at around twice the rate they did between March 2024 and October 2025. This estimate is given with high uncertainty; we would not be surprised if the true acceleration was closer to 1.3x or 3x the historical rate. A majority of this acceleration today is likely attributable to Anthropic's extensive use of AI to do R&D, though there is significant uncertainty and other causes likely played a [...] --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/N5AmYkRks5CqAxvBg/metr-independent-assessment-of-ai-r-and-d-acceleration-at Linkpost URL: https://metr.org/blog/2026-11-05-anthropic-ai-rd-acceleration-assessment/ --- Narrated by TYPE III AUDIO.
November 5 · 22 min
“Efficient Lempel-Ziv Prediction with the Run-Length BWT” by OhadA
Audio note: this post contains a lot of mathematical notation. We narrate the simpler formulas and flag the complex ones, so you may want the original text to hand. The Lempel-Ziv parse of Chad Gadya (חד גדיא), right-to-left, drawn as an arc diagram. Generated by Claude Opus 5.5 I have published a new paper on efficient string prediction. It was written during the PIBBSS 2026 summer fellowship, under the mentorship of Vanessa Kosoy. This research builds on Kosoy's previous work in stringology. Kosoy built a polynomial-time algorithm for predicting repetitive sequences and I have used other algorithmic techniques to further decrease the runtime. The goal of this post is to explain some of the ideas from Kosoy's paper and how my work builds on them. My writing will be mostly informal; the field of string algorithms has a lot of technical jargon which I will not try to fully explain. I hope to mostly communicate interesting ideas and topics I've bumped into during the research, while linking to external resources for anyone who wishes to learn more. The prediction trinity There are three things we want every good predictor to have: [...] --- Outline: (01:29) The prediction trinity (05:23) Matching the string's history (08:23) Going faster (12:37) Going even faster (15:39) Going fastest (20:58) Ending thoughts The original text contained 5 footnotes which were omitted from this narration. --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/voe2dDnuopYn6DC3N/efficient-lempel-ziv-prediction-with-the-run-length-bwt --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.
November 5 · 10 min
“People Should Resolve Tensions In AI Safety Research” by David Africa
By now, there are lots of AI safety research papers, more than what is reasonable for a single person to read. This will get worse as the machines get better at AI safety research, and as more talent and resources pour into the field. But the "paper" is an arbitrary unit, we really care about if we make progress on an "uncertainty" or an "issue", which papers provide heterogeneous and sometimes contradictory evidence towards. Right now, I loosely claim that we should have more short writeups that simply compile evidence from papers (particularly contradictory evidence), and reconcile them into an answer for a question. Why? Object level, it's good for the field. It's hard to sift through all these papers! Organising them is useful and saves time. Orienting towards research questions is a good way to organise our understanding so that we can productively identify gaps and write papers. This was partly what motivated Stress-Testing Alignment Midtraining. Interesting progress comes out of contradictions, where people have differing intuitions and experimental results. Methodological errors can also be exposed if we have two things that should be the same, but are actually different. It's good for field-building reasons. [...] --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/tn4WSyrqr7ozGLjRy/people-should-resolve-tensions-in-ai-safety-research --- Narrated by TYPE III AUDIO.
November 5 · 5 min
“Useful Junk” by jefftk
Occasionally people will get excited about a minimalism fad, where you figure out which of your possessions you need and keep only those. As someone who generally doesn't like getting rid of things, I've been thinking some about why, and the main thing I keep coming back to is that random objects can be helpful in unpredictable ways. Here are some things that I kept without a specific use in mind, but that ended up being useful. You could say they saved time or money, but often the biggest gain was from being able to do a project in one go with what I already had on hand: Ethernet cable. Back in 2019 if you'd asked me whether I would ever use ethernet again, I would have said probably not. WiFi was good enough for everything I was doing. But then I (like a lot of us) suddenly started working from home and got frustrated about how poorly my internet was working for video calls and ran ethernet to my desk. Having cable on hand already meant I could just set it up. Plexiglass. When I built my accidental electronic instrument I wanted [...] --- First published: November 5th, 2026 Source: https://www.lesswrong.com/posts/dYZrtZYty2CSCLjJS/useful-junk --- Narrated by TYPE III AUDIO. --- Images from the article: Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

Showing 41–60 of 598 episodes