Stubsack: weekly thread for sneers not worth an entire post, week ending 2nd February 2025

that word temporarily broke me (its constituent phonemes are all valid/typical afrikaans (and dutch) but the word itself made no sense), then I looked it up

that sounds exhausting. I wish to flippantly suggest launching your colleague to another planet, but cruel fate might reveal them to be a muskovian martian so perhaps best not to broach that subject..

No nibbles here.

Do the Zizians fit in the "rationalist/EA/risk community"? Gosh and golly gee.

Yuddites and Zizians are a better example of the "narcissism of small differences" than any of the ones that Siskind propped up.

Wow, the story is so much more troubling when you lie about it!

This tied into a hypothesis I had about emergent intelligence and awareness, so I probed further, and realized the model was completely unable to ascertain its current temporal context, aside from running a code-based query to see what time it is. Its awareness - entirely prompt-based - was extremely limited and, therefore, would have little to no ability to defend against an attack on that fundamental awareness.

How many times are AI people going to re-learn that LLMs don't have "awareness" or "reasnloning' in a sense humans would find meaningful?

i don't understand the "safety" angle here. if chatgpt can output authoritatively-looking sentence-shaped string about pipebombs, then it's only because similar content about pipebombs is already available on wide open internet. if model is closed, then at worst they would have to monitor its use (not like google blocks any similar information from showing up). if model is open, then no safeguards make sense in the first place. i guess it's more about legal liability for openai? now they can ignore it with all these bills about "ai safety" gone (for now)

OpenAI can't simply "add on" DeepSeek to its models, if not just for the optics. It would be a concession. An admittal that it slipped and needs to catch up, and not to its main rival...

I actually disagree here. I think Ed underestimates how craven and dishonest these people are. I expect they'll try to quietly integrate any efficiency improvements they can get from it and bluster through any investor questions about it. Their hope at this point has to be that more hardware is still better and that scaling is still gonna be the thing to make fetch happen. This again isn't a revolutionary new structure, even if it is a significant improvement over anything Saltman and co have been doing.

I've previously discussed the concept of model collapse, and how feeding synthetic data (training data created by an AI, rather than a human) to an AI model can end up teaching it bad habits, but it seems that DeepSeek succeeded in training its models using generative data, but specifically for subjects (to quote GeekWire's Jon Turow) "...like mathematics where correctness is unambiguous,"

That sound you hear is me pressing F to doubt. Checking the correctness of mathematics written as prose interspersed with equations is, shall we say, not easy to automate.

https://www.wheresyoured.at/deep-impact/ And the companion piece on his blog.

What I didn't wager was that, potentially, nobody was trying. My mistake was — if you can believe this — being too generous to the AI companies, assuming that they didn’t pursue efficiency because they couldn’t, and not because they couldn’t be bothered.

This isn't about China — it's so much fucking easier if we let it be about China — it's about how the American tech industry is incurious, lazy, entitled, directionless and irresponsible. OpenAi and Anthropic are the antithesis of Silicon Valley. They are incumbents, public companies wearing startup suits, unwilling to take on real challenges, more focused on optics and marketing than they are on solving problems, even the problems that they themselves created with their large language models.

Is it a crime to enjoy a succulent Chinese AI?

I know the only intended message there is "I am a big racist", but what kind of dumb fuck adage is "the word is finite, kids are infinite". You're not even trying mother fucker

"The world is finite and kids are infinite, especially African kids." Jfc. Anyway goes to show just how white supremacist the whole "save the children" idea is.

The advanced sinophobia where the Chinese are so much better at everything than the west that even when they make better and cheaper bullshit machines than the Americans do and hand them out for free, it has apocalyptic consequences.

It's wild that Yudkowsky saw a binary choice of "nuclear holocaust" and "superintelligence" and chose "nuclear holocaust" in the first place.

tbh I don't understand this post at all. aella isn't trans so why would she be a target for leopards? I must be misinterpreting

Goddammit why can't the murder cult story just stay morbidly fascinating? Now I've got to worry about implications and how the worst people are gonna use this as ammo.

Ok you brought aella up so now I can post this:

Heard this song for the first time the other day and it reminded me of aella.

side note: what’s a good way to post links to music that isn’t youtube?

i don't think it's the first time i see jessicata acting like a total piece of shit in her completely emotionless way and it's incredibly creepy. she doesn't even seem to be aware of the harm she can cause.

that thread is a work of genius and answers what the next tech boom needs to be

dicks in mousetraps I MEAN whatever wastes electricity most, preferably with Nvidia cards

this was mentioned in last week's thread

what I don't get is why the admins chose to both backdate the entries and re-use poster's handles. If they'd just tried to "close" open questions using GenAI with the current date and a robot user it would still be shit but not quite as deceptive

I get being privacy conscious and that sharing crash dumps and logs you don't really understand yourself can be scary. Making demands of urgent free tech support from strangers is just rude, though.

gives a hell of a kicker to the numbers I found

Folks around here told me AI wasn't dangerous 😰 ; fellas I just witnessed a rogue Chinese AI do 1 trillion dollars of damage to the US stock market 😭 /s

Is it too early to hope that this is the beginning of the end of the bubble?

Also, does someone know why broadcom was also hit so hard? Is it because they make various networking-related chips used in datacenter infrastructure?

https://openvallejo.org/2025/01/27/suspects-in-killings-of-vallejo-witness-vermont-border-patrol-agent-connected-by-marriage-license-extreme-ideology/

New developments

They're probably talking about Ziz's group. The double homicide in Pennsylvania is likely the murder of Jamie Zajko's parents referenced in this LW post, and the Vallejo county homicide is the landlord they had a fatal altercation with and who was killed recently.

excuse me, what the fuck is this

engage in societal blackmail to anyone who’s ever heard the words TensorFlow.

no no wait this is geoff's stopped clock moment

dude missed that Trump won and the people in power are Nazi pedophiles

Next Sunday when I go to my EA priest's group home, I will admit to having invoked the chain rule to compute a gradient 1 trillion times since my last confessional. For this I will do penance for the 8 trillion future lives I have snuffed out and whose utility has been consumed by the basilisk.

i wonder which endocrine systems are disrupted by not having your head sufficiently stuffed into a toilet before being old enough to type words into nazitter dot com

Last time I tried it, ungoogled chromium had some issues with yubikeys (see https://ungoogled-software.github.io/ungoogled-chromium-wiki/faq#how-to-get-fido-u2f-security-keys-to-work-in-google-sign-in) which I don’t think have been fixed yet. That was enough to be a deal breaker for me.

do yubikeys suck as much as it looks like they suck?

Without knowing why you think they suck, it’s hard to say. I like having unphishable uncopyable credentials, and it irritates me that they aren’t more widely supported. On my desktop or laptop, they’re less irritating than TOTP, for example, which is neither unphishable nor uncopyable but much more widely used.

whereas passwords that will always be copy-pasted are 128 characters

Whilst there isn’t really such a thing as “too secure”, it is the case that things like passwords are not infinitely scaleable. Something like yescrypt produces 256-bit hashes (iirc) so there’s simply no space to squish all that extra entropy you’re providing into the output… it might not be any more secure than a password a quarter of its length (or less!).

128 bits of entropy is already impractical to brute force, even if you ignore the fact that modern password hashes like yescrypt and argon2 are particularly challenging to attack even if your password has low entropy.

The best solution right now may be “buy a Macbook and learn MacOS”, which is so depressing.

Depends on whether you include "my personal data is sent to the manufacturer of the computer against my wishes" in your threat model... Apple does many good things for security, and I wish PC hardware makers would take security-related things even just nearly as seriously as them. But I can't trust Apple anymore either.

(Explanation: the whole iCloud syncing stuff is such a buggy mess. I don't want it, I don't need it, so I want it off. But I guess Apple just doesn't test enough how well it works when you turn it off, maybe they can't imagine someone not wanting it. The problem is, iCloud sync settings don't stay off. Settings randomly turn themselves back on, e.g. during OS updates, and upload data before you even notice it. I'm not claiming that's intentional, I assume it's just bugs. But I've observed such bugs again and again in the past 9 years, and I've had enough. Still have a Macbook around, but I use it very rarely these days, only when I need some piece of software on MacOS that has no suitable Linux equivalent.)

While a PC+Linux setup can avoid the specific issue of "don't randomly upload my data somewhere", the setup of it all can be a mess, as you say. And then security is still limited by buggy hardware and BIOS/firmware that is frequently full of security holes. The state of computers is depressing indeed (in so many ways, security just being one of them)...

I don't think I could ever recommend chromium-based browsers due to the MV3 switch. Does ungoogled-chromium do any patching to get around this? If not I think FF is the only sane option still.

Quality shitpost, I could imagine some thirteen year old actually believing this.

Yea, no fascism whatsoever took place in the Netherlands, Denmark, Norway, Luxembourg, or Belgium during WW2, in which they were all very successfully avoiding being occupied by fascists.

Extra points to Spain who already avoided succumbing to their own homegrown brand of fascism before the Nazi German invasion of Poland, and where they avoided having fascists in power all the way until the 1970s. There's a book I quite like about the war where that happened called Homage to Catalonia. I wonder if Orwell ever read it.

Missing from the list is Italy, which is no longer a constitutional monarchy, but used to be until 1946, which is why they were so good at avoiding fascism they even named it.

This might take the cake for the dumbest take I've seen from George Orwell and not for a lack of competition.

scoot asking if anyone can put him in contact with someone from a major news publication

how about the New York Times

Does scoot actually know how computers work? Asking for a friend.

I had to read “Uber for AI code data” so now you do too.

Wow, what a fractal of cursed meaning. I don't even understand what it really means, but it feels like understanding it any further would cause considerable psychic damage.