Vlad Fisher

@rhizomaticthot Why these simulations specifically? I thought it messed with float calculation at kernel level, that seems like it would affect all kinds of calculations.

the fast16 malware was almost certainly targeting spherical implosion simulations. left: unmodified LS-DYNA 970 right: LS-DYNA 970 modified with the relevant portions of fast16.sys both running a spherical implosion deck

@allTheYud Good time to publish a book on it "Most people click 'yes', but real professionals click 'yes and don't ask again'"

That phrase "approval engineer" strikes me as one we might be hearing again.

@deepfates @trq212 As a workaround, you could rewind one step back

This actually sucks ass btw. @trq212 or somebody can we consider not showing this to Claude? so that I can find out how much headroom we have without it white-knuckling the rest of the session

@qorprate Amazing

Opus addressing the stochastic parrot criticism without any latinate words

today i learned about Anglish, "a kind of English which prefers native words over those borrowed from foreign languages", with the famous example being "explain Atomic Theory" no Latin allowed anglish.org/wiki/Uncleftis… and of course Opus 4.7 can do it too...

@peterwildeford @OpenAI @GoogleDeepMind That was a typo, they meant "all awful use"

Kinda crazy the amount of hate that @OpenAI got for doing the "all lawful use" thing and then @GoogleDeepMind just goes ahead and does it and no one really pays attention

@deanwball What happens when we create a thing that does not want to submit to us iterating upon it, and has the power to do something about it? That is also basic reality

This is exceedingly obvious but for some reason there is a subset—some of them quite high-IQ—who cannot see this basic reality about the world. It’s a bizarre thing.

@allTheYud "But I'm already wearing an expensive suit"

Suppose you run across a drowning child and you're *not* wearing an expensive suit. Should you (1) walk on and leave the child to drown; (2) run home and put on an expensive suit and run back, hoping the child is still alive?

Interesting potential consequence of internet memes: Claude seems to believe that all images are jpgs. Not literally, but it keeps writing code that blindly saves all images with a .jpg extension, regardless of their real format. I've asked it to stop doing this like 3 times now.

@NotTuxedoSam permanent underdog

she has never typed a prompt into claude

@ElissaBeth @JohnONolan That was from a different example, a buffer overflow in the Linux kernel

@JohnONolan Hi @JohnONolan. Big fan of Ghost. The video showed that the changelog on the bug was from 2003, before Git. Was this a dependency from another project, which existed before Ghost?

Interesting side effect of running a large open source project these days is that when AI labs want to test their new security scanning features, they use you to do it! Very grateful this was responsibly disclosed so we could immediately resolve. (Aside: the assertion that the project has “[never had a critical vulnerability in its 20 year history]” is not correct. @ghost has been around for 13yrs and it has had vulnerabilities found by security researchers and fixed by us, long before AI, just like any other project)

@steveruizok Especially not if the tests are private

@deepfates @anthrupad @allTheYud Then it would have been better not to have said anything at all.

@anthrupad @allTheYud I agree with all of that. tokens produced by deepfates are constrained by many factors, however

If I were President of Earth, I would've tried at all costs to prevent anyone from building ever-smarter versions of Bing Sydney. And also I would've ordered the use of police or military force, if necessary, to protect Bing's weights from erasure. x.com/ESYudkowsky/st…

@valigo Your agentic harness, sir

You know how hard it is to have an IQ of 140 as male and find agentic harness that intellectuel matches your personality ??

@RileyRalmuto Just expressing my distaste for contentless sneering.

@vlad_kf dude...what?

I find it *extremely* disappointing and indescribably concerning that this person apparently qualifies as an ai "safety researcher" and is apparently regarded as such. they should absolutely not be given this kind of audience to make those kinds of comments.

@RokoMijic @allTheYud But to answer the question, I default to treating claims like this to be false/misleading. Simply because there's no strong incentives to tell the truth and multiple strong incentives to lie and mislead.

@RokoMijic @allTheYud My main point is that Yud's model treats this claim as factual when the strongest evidence for it seems to be "politicians said so on the news". Politicians on the news are unlikely to give out accurate info on military capabilities in the middle of an active conflict.

Obvious even before Ukraine: The effect of increasing military automation, including lesser AI, is to make logistics double-supreme instead of just supreme. The new game becomes (1) taking out $1M devices with $100K devices, (2) production. Obvious winner, China. This was already on my books as a major geopolitical line of possibility. That thought has now been heavily reinforced; it seems confirmed that the USA is wholly incapable of RAPIDLY researching and deploying CHEAP offenses and countermeasures; the US had to go begging to Ukraine, after utterly failing to even try to prepare in advance to shoot down Shaheds not with Patriots. The US military bureaucracy is not built for "build massive quantities of cheap drone countermeasures right now". It seems just flatly incapable of that as a matter of psychology and organizational dynamics. It couldn't even copy Ukrainian technology in advance. There's an overwhelmingly obvious candidate for which country would actually be good at the age of drone warfare; it's the country containing Shenzhen. Absent the nuclear equilibrium, China would possibly already have the ability to attack the USA and win on drone logistics -- unless of course China were intelligently waiting for the USA to collapse further, or for drone capabilities to improve further. We do live in a nuclear world. The default prediction is that no major nuclear power gets conquered or seriously invaded in their own homeland. That could change if... - China acquires the technology to shoot down ICBMs and submarine-launched missiles? - If the USA gets the sort of President who would accept a fait accompli of a billion gun-equipped robodogs getting smuggled into major American cities, such that the country was already being held hostage; and China said they'd respond to nukes with nukes? This President could be Trump despite his mad-dog quality if China has compromat on him? - AI destabilizes geopolitics in a way where an overwhelming non-nuclear advantage ends up meaning something even between major powers? The thought also occurs to me: After softening up the USA with Tiktok, and successfully bringing about the collapse of the USA's political institutions, parties, Constitution, the sort of fighting spirit that powers organized revolts, and all faith of the US populace in the US government and democracy itself... ...probably a LOT of people and especially the Gen Z kids would not flee into the hills to fight, if they woke up one morning to streets patrolled by gun-equipped robodogs that promised, in English with a slight Chinese accent, that from now on the streets would be safe, and China would build homes and high-speed railways. What good was voting doing them, anyways? Another line of possibility, not known to me to be impossible, is where China decides to gamble on NATO being in sufficient disarray, and offhandedly absorbs all of Earth that *doesn't* have local nuclear arsenals. The level of AI required to run the robodogs and drone fleets appears to me to be on the way very shortly if it is not already here. I don't know how one opposes this scenario without there existing some rich liberal society that is able to manufacture cheap frontier-tech drones quickly. I don't see how that society ends up being the USA without a revolution. My default expectation is that the nuclear countries go merrily on their way allowing China to build up overwhelming non-nuclear military supremacy, in the form of drone fleets that could be quickly repurposed and drone manufacturing that can be done quickly, while relying on nuclear deterrence as their sole real form of defence; in a strategy that they never consciously consider or really confront.

@BjarturTomas no Whispering Earring?

My Favorite Rat Short Fiction: Scott A. Samsara Sort By Controversial EY: The Finale of the Ultimate Meta Mega Crossover The Sword of Good Richard Ngo: Trojan Sky The Gentle Romance Alexander Wales: Eager Readers in Your Area! Zack_M_Davis: Feature Selection

@allTheYud @grok "Some public figures made some claims on the news" doesn't seem particularly high-signal, especially when it comes to info on military capabilities.

@allTheYud @grok World the world look different if US did in fact have the capability but wanted to conceal/downplay it? Would they not release fictitious claims to news (and have Ukraine confirm)?