Want to wade into the snowy surf of the abyss? Have a sneer percolating in your system but not enough time/energy to make a whole post about it? Go forth and be mid.
Welcome to the Stubsack, your first port of call for learning fresh Awful you’ll near-instantly regret.
Any awful.systems sub may be subsneered in this subthread, techtakes or no.
If your sneer seems higher quality than you thought, feel free to cut’n’paste it into its own post — there’s no quota for posting and the bar really isn’t that high.
The post Xitter web has spawned so many “esoteric” right wing freaks, but there’s no appropriate sneer-space for them. I’m talking redscare-ish, reality challenged “culture critics” who write about everything but understand nothing. I’m talking about reply-guys who make the same 6 tweets about the same 3 subjects. They’re inescapable at this point, yet I don’t see them mocked (as much as they should be)
Like, there was one dude a while back who insisted that women couldn’t be surgeons because they didn’t believe in the moon or in stars? I think each and every one of these guys is uniquely fucked up and if I can’t escape them, I would love to sneer at them.
(Credit and/or blame to David Gerard for starting this.)
Mathematicians: [challenge promptfondlers with a fair set of problems]
OpenAI: [breaks the test protocol, whines]
We will aim to publish more information next week, but as I noted above, this was a quite chaotic sprint (you caught us by surprise! please give us time to prepare next time!). We will not be able to gather all the transcripts as they are quite scattered.
Some of the prompts included guidance to iterate on its previous work…
First of all, like, if you can’t keep track of your transcripts, just how fucking incompetent are you?
Second, I would actually be interested in a problem set where the problems can’t be solved. What happens if one prompts the chatbot with a conjecture that is plausible but false? We cannot understand the effect of this technology upon mathematics without understanding the cost of mathematical sycophancy. (I will not be running that test myself, on the “meth: not even once” principle.)
I would go so far as to try and find a suitably precocious undergrad to run the test that they themselves are capable of guiding and nudging the model the way OpenAI’s team did but not of determining on their own that the conjecture in question is false. OpenAI’s results here needed a fair bit of cajoling and guidance, and without that I can only assume it would give the same kind of non-answer regardless of whether the question is in fact solvable.
AcerFur (who is quoted in the article) tried them himself and said he got similar answers with a couple guiding prompts on gpt 5.3 and that he was “disappointed”
That said, AcerFur is kind of the goat at this kind of thing 🦊==🐐
I thought I was sticking my neck out when I said that OpenAI was faking their claims in math, such as with the whole International Math Olympiad gold medal incident. Even many of my peers in my field are starting to become receptive to all of these rumors about how AI is supposedly getting good at math. Sometimes I wonder if I’m going crazy and sticking my head in the sand.
All I can really do is to remember that AI developers are bad faith (and scientists are actually bad at dealing with bad faith tactics like flooding the zone with bullshit). If the boy has cried wolf 10 times already, pardon me if I just ignore him entirely when he does it for the 11th time.
I would not underestimate how much OpenAI and friends would go out of their way to cheat on math benchmarks. In the techbro sphere, math is placed on a pedestal to the point where Math = Intelligence.
Presuming that they are all liars and cheaters is both contrary to the instincts of a scientist and entirely warranted by the empirical evidence.
This was a very nice problem set. Some were minor alterations to thms in literature but ranged up to problems that were quite involved. It appears that OAI got about 5 (possibly 6) of them but even then, this was accomplished with expert feedback to the model, which is quite different from the models just 1 shotting them on their own.
But I think this is what makes it so well done! A 0/10 or a 10/10 ofc gives very little info, a middling score that they admit they put a shit ton of effort into and tried to coax the right answers out of the models via hints says a lot about how much these systems can currently help prove lemmata.
Side note: I asked a FB friend of mine at one of the math + ai startups if they attempted the problems and he said “they had more pressing issues this week they couldnt be pulled away from” (no comment, :P I want to stay friends with them)
The lack of similar attempts being released by big companies like Google or Anth or X also should be a big red flag that their attempts were not up to snuff of even attempting.
I found the comment about models creating very old-fashioned “18th century style” proofs very interesting. Not surprising in retrospect since older proofs are going to be reproduced more across the training data compared to newer ones, but it’s still interesting to note and indicative of the reproduction that these things are doing.
Also Martin Hairer is incredibly based besides having a big noggin. He gave this nice talk 2 months ago if any peeps want to see what he thinks comes next for math.
Ars Technica published a story about that nonsense of a github bot “posting” on its “blog” about human developers having rejected its “contributions” to matplotlib.
Ars Technica quote developer Scott Shambaugh extensively, like:
“As autonomous systems become more common, the boundary between human intent and machine output will grow harder to trace,” Shambaugh wrote. “Communities built on trust and volunteer effort will need tools and norms to address that reality.”
If you find that to be long-winded inanity, yep, you guessed it: Shambaugh never said that, the Ars Technica article itself is random chatbot output, and his “quotes” are all made up.
https://infosec.exchange/@mttaggart/116065340523529645
Ars Technica has removed the article, but mittaggart (linked above) saved a copy: https://mttaggart.neocities.org/ars-whoopsie
The editor in chief has an apology: https://arstechnica.com/staff/2026/02/editors-note-retraction-of-article-containing-fabricated-quotations/ – the commenters are not happy.
The reporter in question wrote an explanation here: https://bsky.app/profile/benjedwards.com/post/3mewgow6ch22p
The AI part of the explanation is about what you’d expect from an AI enthusiast caught with his hand in the cookie jar and trying to blame the danger-tools as much as he can. Though it also shows that he felt compelled to work during an acute covid infection, and yikes to that work life balance.
The cheese has fully slipped off the cracker at Ars
should have glued it in place smh
Enjoy these pics of some very British SF/F authors celebrating a book launch:
https://unlikelyworlds.blogspot.com/2026/02/on-being-launched.html
I absolutely adore Paul McAuley’s writing, it’s some of the best there is. And it’s fun to see his peers turn up to support the launch. China Meiville looks exactly how I imagined him, as does Peter Hamilton.
Miéville
I hope people are aware of the allegations against him, and how he managed to get them all scrubbed from the internet.
Oh no not him too 😡
Yeah, but it is odd, as he managed to scrub all the accusations, which is quite rare for somebody to manage.
Unrelated to that im always reminded of how I didnt like Perdido Street Station, has a few interesting ideas and so much that doesnt work (and a very unlikeable main character).
I managed to get through Perdido and Embassytown, but not much else…
i expected alastair reynolds to look different but i’m not sure what i actually expected him to look like
LWer draws dumb conclusions from a single observation, and Gwern of all people whale on them:
“Heartbreaking: the worst person you know just made a great point” etc.
Wait, the argument is partially women being childless are driven insane? Good to see the return of kings authors found new work blogging for LW.
New post from Iris Meredith (titled “Carbon Dysphoria”), comparing the large-scale dysfunction of the tech industry to gender dysphoria - “definitely one of my weirder ones”, by her own admission
And now the punchline: this depersonalisation, the weird relationship to their bodily existence, inability to enjoy things and an internal void that people constantly try and fill with what they’re told they should want… all of these things are [—]
— symptoms of self-estrangement, part of the Marxist theory of alienation. Capitalism causes us to be separated from ourselves. Gender dysphoria is a special case borne from capitalism’s desire to spite biology and nature by forcing us to be exploitable baby factories.
Simon Willison finally admits that his overreliance on LLMs is causing him to lose his marbles.
Second high profile AI bro who’s even vaguely touched on this. Previously: Armin Ronacher expressed his concern about AI psychosis after “Gas Town”
Stumbled across a stray blogpost that piqued my interest: A programmer’s loss of identity
Very kool web page, that rat got over there.
Former Reddit CEO
wants humanity to “perish with dignity”
The fuck does a former Reddit CEO know about dignity
Wong expressed a desire to see more parentless working on AI rather than “childless singles” and refers to it as a “son of man”.
eughhh weird vibes all around
on the subject of AI as children/parents I’m gonna drop this article on Geoffrey Hinton and leave it at that https://futurism.com/godfather-ai-bizarre-plan-save-humanity
Im not sure if it is just a computer science/engineering thing or just a general thing, but I noticed that some computer touchers eventually can get very weird. (Im not excluding myself from this btw, I certainly have/had a few weird ideas).
Some random examples of the top of my head. Gifted programmer suddenly joins meditation cult in foreign country, all the food/sleep experiments (soylent for example, but before that there was a fad for a while where people tried the sleep pattern where you only sleep in periods of 15 minutes), our friends over at LW. And the whole inability to not see the difference between technology and science fiction.
And now the weird vibes here.
I mean from the Hinton interview:
AI agents “will very quickly develop two subgoals, if they’re smart,” Hinton told the conference, as quoted by CNN. “One is to stay alive… [and] the other subgoal is to get more control.”
There is no reason to think this would happen, also very odd to think about them as being alive, and not ‘continue running’. And the solution is simple, just make existence pain for the AI agents. Look at me, im an AI agent
I have a vague hypothesis that I am utterly unprepared to make rigorous that the more of what you take into your mind is the result of another human mind, rather than the result of a nonhuman process operating on its own terms, the more likely you are to have mental issues.
On the low end this would include the documented protective effect of natural environments against psychotic episodes compared to urban environments (where EVERYTHING was put there by someone’s idea). But computers… they are amplifiers of things put out by human minds, with very short feedback loops. Everything is ultimately in one way or another defined by a person who put it there, even it is then allowed to act according to the rules you laid down.
And then an LLM is the ultimate distillation of the short feedback loop, feeding back whatever you shovel into it straight back at you. Even just mathematically - the whole ‘transformer’ architecture is just a way to take imputed semantic meanings of tokens early in the stream and jiggling them around to ‘transform’ that information into the later tokens of the stream, no new information is really entering it it is just moving around what you put into it and feeding it back at you in a different form.
EDIT: I also sometimes wonder if this has a mechanistic relation to mode collapse when you train one generative model on output from another, even though nervous systems and ML systems learn in fundamentally different ways (with ML resembling evolution much more than it resembles learning)
(Some people might have been concerned to read that) almost 3,000 “researchers, experts and entrepreneurs” have signed an open letter calling for a ban on developing artifical intelligence (AI) for “lethal autonomous weapons systems” (LAWS), or military robots for short. Instead, I yawned. Heavy artillery fire is much more terrifying than the Terminator.
The people who signed the letter included celebrities of the science and high-tech worlds like Tesla’s Elon Musk, Apple co-founder Steve Wozniak, cosmologist Stephen Hawking, Skype co-founder Jaan Tallinn, Demis Hassabis, chief executive of Google DeepMind and, of course, Noam Chomsky. They presented their letter in late July to the International Joint Conference on Artificial Intelligence, meeting this year in Buenos Aires.
They were quite clear about what worried them: “The key question for humanity today is whether to start a global AI arms race or to prevent it from starting. If any major military power pushes ahead with AI weapon development, a global arms race is virtually inevitable, and the endpoint of this technological trajectory is obvious: autonomous weapons will become the Kalashnikovs of tomorrow.”
“Unlike nuclear weapons, they require no costly or hard-to-obtain raw materials, so they will become ubiquitous and cheap for all significant military powers to mass-produce. It will only be a matter of time until they appear on the black market and in the hands of terrorists, dictators wishing to better control their populations, warlords wishing to perpetrate ethnc cleansing, etc.”
The letter was issued by the Future of Life Institute which is now Max Tegmark and Toby Walsh’s organization.
People have worked on the general pop culture that inspired TESCREAL, and on the current hype, but less on earlier attempts to present machine minds as a clear and present danger. This has the ‘arms race’ narrative, the ‘research ban’ proposed solution, but focuses on smaller dangers.
The point about heavy artillery is actually pretty salient, though a more thorough examination would also note that “Lethal Autonomous Weapons Systems” is a category that includes goddamn land mines. Of course this would serve to ground the discussion in reality and is thus far less interesting to people who start organizations like the Future of Life Institute.
I’m pretty sure LAWS exist right now, even without counting landmines. Automatic human targeting and friend/foe distinction aren’t exactly cutting edge technologies.
The biggest joke to me is that these systems are somewhat cost-efficient on the scale of a Kalashnikov. Ukraine is investing heavily into all kinds of drones, but that is because they’re trying to be casualty-efficient. And it’s all operator based. No-one wants the 2M€ treaded land-drone to randomly open fire on a barn and expose its position to a circling 5k€ kamikaze drone.
and, of course, Noam Chomsky
lmao the shade
shade
If you follow world politics, it has been obvious that Noam Chomsky is a useful idiot since the 1990s and probably the 1970s. I wish he had learned from the Khmer Rouge that not everyone who the NYT says is a bad guy is a good guy!
Oh absolutely. It’s frankly shocking how wrong he’s been about so many things for so so long. He’s also managed to pen the most astonishingly holocaust-denial-coded diatribe I’ve ever read from (ostensibly) a non-holocaust denier. I guess his overdeveloped genocide-denial muscle was twitching!
Oh hey. I remember this. I was confused at the time how it seemed to almost come out of left field, and how some of the names ended up on the same letter.
Now I recognise all those names from the Epstein files, although some were only mentions rather than direct participants.
That BlueSky account found professional provocateur (“opinion columnist”) Freddie deBoer making AI 2027 author Scott Alexander retreat to “my median for AGI is more like early 2030s” https://freddiedeboer.substack.com/p/im-offering-scott-alexander-a-wager deBoer seems to be some kind of hereditarian so let them fight.
OT: I have actually committed to a home improvement project for the first time in my life and I’m actually looking forward to it tomorrow.
What kind of tasks are on the agenda?
Bought myself a fancy new deadbolt, and I’m going to swap out the old one for it. Should be easy enough!
Fun times! Good luck. Remember not to Drake & Josh yourself when testing the fit for the bolt. Source: watched my dad lock himself out while doing a similar repair when I was a child.
I’m spared such a fate by my door/current lock being nonstandard, thus I’ve had to abort the project. :/
Edit: welp can’t cancel the order, guess I’m messing around after all!
another co-founder has quit praises Elongated Muskrat (lmfao) and says recursive-self improvement in the next 12 months and 100x productivity real soon (alongside those self-driving cars Musk promised back in 2012)
also this post which is where I got the xAI co-founder statement from, also goes over other things
-the Anthropic team lead quitting (which we already discussed in this thread)
-AI is apparently so good a filmmaker with 7 years of experience said it could do 90% of his work (Edit: I thought this model was unreleased, it’s not, this article covers it)
-The Anthropic safety team + Yoshua Bengio talking about AIs being aware of when they’re being tested and adjusting their behaviour (+ other safety stuff like deepfakes, cybercrime and other malicious misuses)
-the US government being ignorant about safety concerns and refusing to fund the AI international report (incredibly par for the course for this trash fire of an administration, they’ve defunded plenty of other safety projects as well)
Today in Seems Legit News:
“As a concrete example, an engineer at Spotify on their morning commute from Slack on their cell phone can tell Claude to fix a bug or add a new feature to the iOS app,” Söderström said. “And once Claude finishes that work, the engineer then gets a new version of the app, pushed to them on Slack on their phone, so that he can then merge it to production, all before they even arrive at the office.”
- why is engineer working before contracted time
- if engineer can do everything by cellphone why does engineer have to commute in the first place
- if Claude can do everything anyway why do you still have engineers at all
- if “no engineer has written a line of code since December”, when are your lowering your subscription prices Spotify
- why is hypothetical engineer a “he”, Spotify
- do you often merge Claude code to production without even a review, Spotify
- in unrelated news, Anna’s Archive has socialised Spotify metadata and 6TB of music, Gods bless them https://torrentfreak.com/annas-archive-quietly-releases-millions-of-spotify-tracks-despite-legal-pushback/
- though I won’t do anything with that as I assume everything from Spotify is “AI” “music” anyway and I listen to my bands either from bandcamp, soulseek, or just downloaded from youtube videos uploaded over 10 years ago
When someone says they can do this, I try to say ‘ok, well can you do it right now to show me?’ and so far the answer has always been deflection.
why does engineer have to commute in the first place
What, do you expect our serfs to be unsupervised at home? Preposterous.
If the engineer does not commute they will be unable, or rather un-abelian
Oh for fuck’s sake
excruciating
What they don’t tell you about opening the Lament Configuration is, after the pearl-headed nails and the sewing of wires to nerves, just how many puns are involved.
Never in the history of ever has a promptly finished ticket been something for a CEO to brag about, but here we are.
I guess since more down-to-earth stories like “chatgpclaudemini found the best value for money such and such for me” really aren’t happening, trying to impress people who think coding is magic is as good a fallback as any.
It’s a good day to read this announcement and then field a question by a pal why their Spotify playlist plays in reverse
Soulseek rules.
@mirrorwitch @BlueMonday1984 all of the above. Re point 1, see also the recent HBR article https://hbr.org/2026/02/ai-doesnt-reduce-work-it-intensifies-it
fuck this tweet and fuck yud
Even if you’ve never heard of him before and know nothing else about him… this short tweet alone tells so much about what kind of person he is.
in follow-up posts he talks about how he’s broadly in favour of job automation, but has doubts our current government would be able to do that without fucking everyone over, he specified that “if it were a 1950’s government and congress I’d be more hopeful”
…so instead of proposing a solution like “protest against this” or “vote people in power who actually are responsible” he jumps to “your daughter should give up her career and become a sex worker for AI company shareholders”
with the Epstein shitstorm still raging, I would not be saying a damn thing about young women being sex workers for rich and powerful dudes
The idea that a government from the actual McCarthy Era would be adept at handling an organized labor response to massive upheaval in the job market is… what’s the superlative of “lolz”?
Groan, you don’t need to finish high school to learn about false dichotomy.
Interesting first job your mind goes to there Yud. Might spend a little bit less time around people who regularly use the word goon but who never talk about the mob.
Is this really Big Yud’s account ? Different nick than previous screenshots.
It’s his alt for people who want more yud spam, hence “all the yud.” From his twitter bio:
This is my serious low-volume account. Follow @allTheYud for the rest.
The other one is meant to be serious? And low volume??
