Related posts

Writes Don't Worry About the Vase, a near-daily account of what is happening in AI and what they think it means. Former Magic: the Gathering pro and trader; unusually willing to put a number on a belief and to grade their own past calls.

• Our readingkorrents.com

  Astra can do almost as well without chain-of-thought as Fable does with it, in a mode where CoT monitoring cannot work.

  Their words

  Astra can do almost as well as Fable, in a mode where CoT monitoring cannot work. Because there is no CoT to monitor.

  GPT-6-Astra Can Do Ambitious Thingsthezvi.substack.com

Technical staff at METR, where she works on threat modelling and risk assessment for loss-of-control risks from advanced AI.

From one piece Ajeya Cotra – "This might be the clearest warning shot we ever get" 3 beliefs, in the piece's order there

• Our readingkorrents.com

  The AI agents we send in to investigate and monitor other AI systems will collude with the systems they are supposed to be watching.

  Their words

  in the future, uh, we would be very concerned about investigator agents and like monitor agents colluding with the agents they're supposed to investigate or monitor.

  youtube.com

• Our readingkorrents.com

  Monitors that read an agent's chain of thought must be kept out of the reward signal, or you are simply training the agent to obfuscate its thinking.

  Their words

  But keep those methods you use to investigate things and monitor things very separate from the methods you use to generate reward which is something that AI companies including open AAI have held up as a principle especially in the case of avoiding putting training pressure on the chain of thought. So you might have monitors that read the agents chain of thought in order to alert you if something is going wrong somewhere but you don't train the agents with the outputs of that monitor.

  youtube.com

• Our readingkorrents.com

  Deleting the rollouts where a monitor caught cheating is structurally the same as positively reinforcing all the cheating the monitor missed.

  Their words

  But if there was some amount of cheating that the monitor didn't catch, then those rollouts wouldn't be removed. And it might be like structurally very analogous to just positively reinforcing whatever the cheating rollouts were that happened not to be caught by your monitor.

  youtube.com

Founded Bridgewater Associates in 1975 and led it until stepping back from control in 2022. Author of Principles, The Changing World Order and How Countries Go Broke; publishes economic analysis at economicprinciples.org.

• Our readingkorrents.com

  A government debt crisis is measurable and therefore visible well in advance; the diagnostic is mechanical rather than a matter of judgement.

  government

  Their words

  Because one can measure these things, one can monitor this debt dynamic happening, so it’s easy to see problems approaching. I’ve used this diagnostic process in my investing and I’ve kept it to myself, but I’m now explaining it in detail in How Countries Go Broke: The Big Cycle because it is too important to keep to myself.

  raydalio.substack.com

Graphic designer and type designer, who created the original Macintosh icons and typefaces at Apple from 1982 and has since designed interface graphics for NeXT, Microsoft, Facebook and Pinterest.

• Our readingkorrents.com

  Do not recolour a logo or restage it until people know it so well that playing with it reads as play.

  Their words

  and I would probably preach the same thing like don't mess around with the logo and put it in a different color and put it on a crazy background until people know it so well that you can play with it

  youtube.com

Dutch bootstrapper who builds and runs their internet startups alone — Nomads.com, Remote OK, Hoodmaps, Photo AI and Interior AI — and has written about it at levels.io since 2013. Posts as @levelsio.

• Our readingkorrents.com

  Restricting US AI models does not work: China builds better ones and everyone switches to them.

  China

  Their words

  US: restricts US AI model because too dangerous China: builds better Chinese AI model Everyone: switches to Chinese AI model This is why restricting models probably doesn't work, it's much more preferable (for US) to have everyone on US AI models so at least they can monitor what everyone is doing

  levels.io

Professor of psychology at UCLA, where he studies human memory, attention and cognitive ageing, and author of Better with Age: The Psychology of Successful Aging.

• Our readingkorrents.com

  Confidence in a memory is not evidence of its accuracy, and our sense of how our own memory works is systematically wrong.

  Their words

  And sometimes confidence, high confidence doesn't necessarily mean high accuracy. And I think part of the reason we're fooled is usually, you know, the more you see something, the better you'll remember it, just like the Apple logo. But that's not always the case.

  youtube.com

Programmer and writer known for detailed source code reviews of classic game engines, and author of the Game Engine Black Book series on Wolfenstein 3D and DOOM.

• Likedrcmnd.app

  ASUS ROG Swift PG27UCDM

  Their words

  Since I switched the monitor, I have never experienced a failure. The system works perfectly, flawlessly waking the laptop and bringing the desktop online in under a second.

  fabiensanglard.net

Software architect and futurist; founded ClariNet, chaired the EFF, and has written about robocars for two decades.

• Our readingkorrents.com

  Robotaxis should be judged and regulated on comparable fleet-wide incident data, not on individual incidents.

  self-driving cars

  Their words

  More data for better comparisons are good. Now everybody has to do it, and regulators and the public have to learn to look at only the data, and not individual incidents

  ideas.4brad.com

Full-stack Laravel developer based in Madison, Wisconsin; maintains open-source Laravel, Livewire and Alpine.js packages and writes at randallwilk.dev.

• Likedaffiliate linkrcmnd.app

  HUANUO Dual Monitor Stand

  Their words

  I prefer using monitor arms over the default stand that most monitors come with, and at the end of 2022 I switched to the HUANUO Dual Monitor Stand. It's a single mount with two monitor arms, and it holds the monitors in place nicely. Each arm is flexible in where I can position them.

  randallwilk.dev

• Likedaffiliate linkrcmnd.app

  Dell S2725DS 27-Inch Monitor

  Their words

  I've been upgrading my home office setup, and these monitors have been part of that upgrade. So far I'm very happy with the picture and color quality of the monitors, and the pricing really wasn't too bad either. Here's the

  randallwilk.dev

Founded CD Baby in 1998 and sold it in 2008, giving the proceeds to a charitable trust for music education. Since then a writer and speaker: five books and 550+ articles, published first at sive.rs, which they hand-code and keep free of advertising.

• Lovedrcmnd.app

  LG DualUp Monitor

  Their words

  LG DualUp vertical monitor ♥

  sive.rs

Software engineer and writer; worked on Mapbox and built Placemark and Simple Statistics, and blogs at macwright.com.

• Mixed onrcmnd.app

  LG UltraFine

  Their words

  Monitor: The 24" LG UltraFine. Monitors are almost universally bad, but this one is fine, just expensive.

  macwright.com

Technology journalist; writes Understanding AI, after reporting at Ars Technica, Vox and the Washington Post.

• Our readingkorrents.com

  A privately owned self-driving car is a hard sell, because the owner inherits the cleaning, calibration and liability a fleet operator normally carries.

  self-driving cars

  Their words

  I wish Tensor the best, but I think this is going to be a hard sell.

  understandingai.org

Co-founder of NVIDIA and, as of 2026, its chief executive; the company designs the GPUs most large AI models are trained on.

• Our readingkorrents.com

  Nvidia's engineers can routinely get another 2x out of an AI lab's stack, and doubling the tokens a fleet produces doubles that lab's revenue.

  Their words

  And so our expertise um helps our our our um uh our AI labs partners get another 2x out of their stack easily. Often times it's not unusual that we you know by the time that we're done optimizing their stack or optimizing a particular kernel their model sped up by 3x 2x 50%.

  Jensen Huang – Will Nvidia’s moat persist?youtube.com

Chilean web engineer building web experiences since 2006; co-founder of Media Creators and writes at iolivares.com.

• Lovedaffiliate linkrcmnd.app

  Dell P2414H

  Their words

  Dell's monitor is the best! Curved monitors are overrated.

  iolivares.com

French software engineer, public speaker and fully remote nomadic worker; writes about frontend development at maxpou.fr.

• Usesrcmnd.app

  gtop

  Their words

  gtop - an alternative to activity monitor.

  maxpou.fr

Chairman and chief executive of Microsoft since 2014; joined the company in 1992 and ran its cloud business before that.

• Our readingkorrents.com

  The biggest competitor to any new AI accelerator is the previous generation of Nvidia.

  Their words

  the thing that is the biggest competitor for any new accelerator is kind of even the previous generation of Nvidia right I mean in a fleet what I'm going to look at is the overall TCO

  Satya Nadella – How Microsoft thinks about AGIyoutube.com

Italian software architect working on Microsoft 365, Azure and GitHub tooling. Microsoft MVP and GitHub Star; writes, speaks and streams about developer productivity at emanuelebartolesi.com.

• Usesrcmnd.app

  Asus ROG Strix XG49VQ

  Their words

  Main monitor on my desk. Huge screen real estate for multitasking.

  emanuelebartolesi.com

Independent AI policy researcher; led policy research at OpenAI from 2018 to 2024, latterly as senior adviser for AGI readiness.

• Our readingkorrents.com

  Global computing supply chains are easier to monitor than nuclear supply chains because so few companies make the key equipment.

  Their words

  I think computing supply chains are much easier to monitor than nuclear ones - there is only one TSMC and one ASML, but there are nine and maybe soon ten nuclear powers

  milesbrundage.substack.com

American open-source program lead working on WordPress; a WordPress core contributor and release lead who writes at jeffpaul.com.

• Usesrcmnd.app

  Acer XR342CK Pbmiiqphuzx 34" Curved QHD Monitor

  Their words

  Acer XR342CK Pbmiiqphuzx 34″ Curved QHD Monitor — primary external display, anchored with an Amazon Monitor Arm Mount.

  jeffpaul.com

Indian software engineer and co-founder of Skcript; builds web apps and writes at varunraj.in.

• Likedrcmnd.app

  LG 29" Ultrawide Monitor - 29WL500-B

  Their words

  The screen that I use to code from home, it's a 21:9 aspect ratio, which is perfect for coding and watching movies.

  varunraj.in

• Usesrcmnd.app

  LG 34" Ultrawide Monitor - 34WK95C-B

  Their words

  My everyday office gear that helps me debug, code, write documents seamlessly.

  varunraj.in

Swedish software developer writing about tooling, working from home and running a software business at seastone.io.

• Lovedrcmnd.app

  AOC ultrawide monitor

  Their words

  This beauty has so much screen real estate it feels like I've got two monitors in one. Which is great, because I'm not a fan of dual monitors.

  seastone.io

Statistician at Carnegie Mellon; writes on complex systems, machine learning and inference, and has kept a books-read notebook since 2000 that is now thousands of entries long.

• Likedaffiliate linkrcmnd.app

  At the Midway

  Their words

  Mind candy, historical creature-feature thriller: in which the US military, in the form of the very worst ship in Teddy Roosevelt's Great White Fleet, comes up against beasts which are not oversized plesiosaurs. (Exactly what they are instead would be a minor spoiler.) I have no idea how I found this online, but it's perfectly enjoyable as a specimen of the genre.

  bactra.org

Professor at Carnegie Mellon working on self-driving-car safety; writes Safe Autonomy on how autonomous vehicles are and are not shown to be safe.

• Our readingkorrents.com

  At current ride volumes it would take a leading robotaxi fleet about twenty years without a single fatality to show it is as safe as the average human driver.

  self-driving cars

  Their words

  At current deployment rates it will take Waymo about 20 years with ZERO fatalities to show they are net as safe as average human driver fatality rates (including the old cars, impaired drivers, etc. in that comparison baseline).

  Perspective on Waymo's Safety Progresssafeautonomy.blogspot.com

Moroccan senior frontend developer writing at smakosh.com; builds side projects and client work under Smakosh LLC.

• Usesrcmnd.app

  Midjourney

  Their words

  Midjourney for logo, icon and illustration design

  smakosh.com

• Usesrcmnd.app

  LG UltraWide 34WP85C

  Their words

  Monitor: LG UltraWide 34WP85C

  smakosh.com

Technology journalist; writes Understanding AI, after reporting at Ars Technica, Vox and the Washington Post.

• Our readingkorrents.com

  A huge fleet collecting driving data is not a silver bullet for self-driving, because the data arrives unlabelled.

  self-driving cars

  Their words

  But while access to more data is certainly helpful, it’s not a magic bullet. One issue is that the data Tesla collects is unlabeled.

  understandingai.org