Welcome to November, and 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 soo 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.)
All the US markets are down today. Gary Marcus is declaring the GenAI bubble burst - I think that’s a little premature but I am smiling seeing Coreweave at -55% of their value at the start of the year.
I don’t think the bubble has truly burst until OpenAI and/or Anthropic go down, but I do however think that this (and the blue wave at the elections) marks that time is running out for the AI industry
I will not say the bubble has burst until OpenAI and Anthropic are dead.
todays adexchanger had a lolworthy note that the binance hack began with a search ad.
If you hadn’t heard, Binance was totally pwned in a $6 billion crypto heist last week, suspected to be a North Korean operation. Well, Binance has now published what amounts to a preliminary autopsy report of the incident.
The heist began when a Binance Wallet DevOps team member accidentally downloaded a malicious copy of Homebrew – a popular open-source tool – from a mirror site promoted in a search ad. The malware was spotted, but not before the hackers had obtained the real identities of Wallet DevOps team members.
Using those identities, the North Korean hacking unit spent months targeting Binance’s wallet team with covert schemes. One involved a prolonged fake job interview that lured a Binance DevOps engineer into downloading an “assignment” onto their work device.
After taking control of the Binance DevOps engineer’s laptop, the hacker used an unnamed internal AI tool to log in to the devices of other engineers. Access to multiple credentials allowed the heist to go through.
Binance guy fell for “top link is ads”. COME ON.
Never understood why my employer would drop $30m on Crowdstrike but not add uBlock Origin to their company chrome profile for free.
I hope they’re all pissing their pants right now
To be fair, who could have predicted that fascism would be unpopular and unstable?
They banked on Trump & co being reasonable competent at fascism and getting rid of meaningful democratic participation in the 2 years before the midterms. And if he’d had a bit more of a plan apart from building marble and gold monstrosities, he’d have gotten away with it, too!
That’s the thing about fascism: incompetence and corruption are built in!
Many tech figures have spent the past few years relocating operations [to Texas], including former trillionaire Elon Musk.
Emphasis mine
Zing 🔥🔥🔥
thoughts & prayers
Any word on if Yarvin is running to Thiel’s south american mansion yet?
Okay yet another post about Yud’s book and its website but finally managed to find that one page I was looking for again. They only put one computer science expert in the praise category shockingly enough (two if you count Scott Aaronson)
Also the main page appears to endorse Dwarkesh’s “AI civilisations” narrative and buys the Claude Mythos hype (its under a read more)
Bart Selman (the other computer scientist) might be rationalist aligned. there’s a future of life institute article on him
Found another ‘interesting’ website: aidoomsdaycountdown.com
one of the options predicts AGI just barely by the end of the year. lmfaooooo
The website puts expert consensus at 78% but does not disclose who those experts are. I have a PhD in computer science and natural language processing and I think this is horse shit. As do most of my peers and my PhD supervisor/prof who I still talk to. I don’t remember being asked and pretty sure my network weren’t either.
I suspect:
We got 10 of my EA friends together. Seven of them agreed and one of them got a bit waffly so I put them down for the 8%.
The website admits as much:
We asked the smartest AI systems on Earth when they think AGI will happen. Their answers might surprise you.
“The smartest minds agree: NFTs are about to change everything. Here’s exactly when it happens”
@pikesley @lurker @BlueMonday1984 that nft thing aged like a milk 😭
Countdowns given to the second, but which only differ by the number of days. How nice of the AGIs to all wait until 00:00:00 UTC!
Clearly written by someone who spends more time chatting with chatbots than talking to actual people. The bulk of the argument is that he asked the chatbots when AI was gonna foom and they told him “In a few years I guess my man”.
The “reading progress” widget bugged out for me and got stuck at 92%, it also links to a page that doesn’t exist.
Still curious? We’ve answered 50+ more questions. […] Showing 22 of 22 questions
I was promised 50+ FAQ questions :(
Anyway I stumbled across the arbitration clause so I’d better stop looking at the website in case I ever have to sue this guy for any reason!
Vibe coded websites really are the epitome of AI: there’s tons of flashy stuff that looks impressive at first glance, but shit starts breaking down with the slightest breeze. The reading progress widget was working perfectly for me at first, but then I clicked around a bit (the “see what’s coming” link in that widget straight up leads to a 404) and then it got stuck at 88% for me. The slider in the “Timeline Explorer” is also extremely laggy for me, with an update taking 1-2 seconds. There’s also some UI elements that don’t make sense, like the four options in the bar above the “Countdown to AGI Achievement” box are duplicated in the “AGI scenarios” sidebar on the right side immediately below. Sometimes the drop down menus from the top just open randomly when I’m clicking something else. Also some buttons break occasionally and I need to refresh to fix them.
Even though my own work does not involve any coding, I have seen a vibe coded project secondhand from someone who wanted to use it because they thought it would save time. It is very impressive at the start, since it can just pump out tons of features very quickly. But then disillusionment sets in as the project turns into a game of whack-a-mole where you yell at the AI to fix an issue and then two more pop up. And no way in hell are you fixing things yourself, because the codebase is an unreadable pile of spaghetti.
Yeah a lot of the text pinged my ‘this was AI generated’ brain
What on Earth do these percentages mean? I scrolled all the way to the bottom and I am none the wiser. No obvious hints in the DOM either. Maybe their chatbot told them it sums to 100%?
I’d wager its the likelihood of said prediction being true
That would be quite the slopbro’s take on probability. Time is continuous, and every fool knows the probability of a continuous variable being equal to something is zero. You can give a probability for it being less than some value, or greater, or within some other range… but none of the percentages relate to the dates in any consistent way. For example, it couldn’t be 30% before 2035 because then you couldn’t possibly have 50% before 2029. But if you just treat it as categorical, as probabilities for safety and risk… then it still doesn’t work, because the categories are all mutually exclusive.
if rationalists knew about error bars, they wouldn’t multiply numbers pulled from their asses the entire time, and what else could they do then
candlestick charts X matrix mult
just imagine what cursed nonsense they might go for
@lurker @BlueMonday1984 Well I’m convinced.
the midterms made Scammy A loopy
"OpenAI is against the notion that it would be acceptable for people to be relegated to a role of doing things that ‘AI systems think makes sense to reward in humans,'” he wrote. “People should not cede control of the future to superintelligence, and that requirement should ultimately constrain AI progress.”
I hope he’s pissing his pants.
Looks like Brodie Robertson (one of the many weird Linux adjacent Youtubers) isn’t able to get any guests for his podcast anymore. Every episode since he invited DHH is a solo episode.
wait oh god there’s more than just Lunduke???
He plays the enlightened centrist and keeps platforming fascists, racists and homophobes. His fan club is also notorious for harassing developers.
just a quick note that our very favourite mr. musk spent half a bil in us dollars on election shenanigans and all he got is blue texas, and i mean it, texas is now blue across the whole slate, including the governor and the u.s. senator.
the very effective mr. musk!
it was quite funny how, while counts were going up, he just entirely shut the fuck up for about 3~4 hours. not a tweet in sight
must’ve been tragically hard overcome that rage. poor pissbaby, so sad. anyway,
Although my district wasn’t very contested, I am still happy that I voted. For the past two years, the oligarchs have operated under the assumption that they are immune from consequences. The AI bros believed that politics was an irrelevant pastime beneath their notice. If anything, I am happy knowing that they now feel fear.
I have more hope for putting serious pressure on my middle-of-the-road Dem rep. The district here only really turned Dem around 20 years ago thanks to the influx of Bay Area bedroom-community commuters, but there are still plenty of families around who immigrated as ag workers. So he already felt a degree of pressure, but now I feel like urging him to get on board with properly punishing the very worst Trump excesses might (might!) not be an empty gesture anymore.
Does that mean Musk is now legally obligated to transition or was it just the children?
Trans people continue to be very worried about US politics despite the blue wave. In Washington and Colorado the trans sports ban is ahead. In Colorado the trans youth healthcare ban is ahead. A lot of the transphobic democrats are doing OK vote wise.
The very real risk here is that trans rights continue to backslide even under a democratic congress or, two years from now, a democratic congress and executive branch.
So while the transgender for everyone memes are appreciated, we’re not out of the woods yet. Liberals are also susceptible to anti-trans propaganda and boy howdy there has been a lot of it.
All that said: the trans council does not bestow the honor of transition on Musk. He must remain cis and miserable.
While I have absolutely no hope in democrats or liberals for meaningfully supporting trans people, I would love to be proven wrong.
I am at least glad that hunting down trans people and burning down any place where people (especially kids) can get gender affirming care is probably not going to be an active priority up and down the various levels and branches of government the way it seemed to be for a while there. I’m not entirely convinced it’s a step in the right direction yet, but it should at least be a temporary halt in most areas.
This is a pattern, with the exceptions of AFD and Trump, every time he supports a guy they lose. The guy has the moneybags of death.
Anyway well done US voters. Yall won a decisive battle now win the war.
@Soyweiser @mawhrin his principal problem is that he’s an utterly repellent loser and everybody hates him 🤷
Yep, and no amount of building toys can change that. Lets hope blue Texas is going to take taxing him a bit more seriously.
hopefully also a lot of accountability and secrets opened up
congratulations to the us voters for the blue tidal wave!!
Also, with such a big blue wave theres no way in hell JD Vance is becoming president like AI 2027 predicted (that was already suspect before all this)
The rationalist attitude towards politics really is something else. Either politics is a pointless sideshow compared to the machine god, or they are the ultimate politics understanders. I could not come up with a worse way to market AI than they did.
I’m not the most politically astute person, but even I know that AI is a perfect symbol of the hated oligarchy. AI is a tool for the rich to plunder the labor and resources of the poor, all while dehumanizing them in favor of the machine. Of course AI is politically toxic. The rationalists are so far up their own asses that they can’t even think of this basic viewpoint.
I used to dislike when Bernie Sanders would take the warnings of the AI doom prophets too seriously, but now I realize that it is still a shrewd strategy, because the outcome is good in the end (bad for the AI companies). If they start talking about how AI will cause human extinction, it is bad either way if they are sincere or if they are grifting. Jon Ossoff even referenced the human extinction narrative in his victory speech. The political reaction was not “please help the AI companies figure out AI safety,” but “shut that shit down right now.” And that is the right thing to do.
or they are the ultimate politics understanders.
A lot of them fail to grasp how rightwing they (rationalists) actually are. Others have ideas that are extremely simplistic, like there are lots of lesswrong post proposing one cool trick to get trump on their side or get normies to understand AI doomerism.
The political reaction was not “please help the AI companies figure out AI safety,” but “shut that shit down right now.”
Yeah, they have all these big brain plans for research programs into aligning AI, not getting that if they actually get the public or any major political figures to really believe their doomer ideas “shut that shit down” with maybe a side of “throw some people in jail” is the common sense response.
A lot of them fail to grasp how rightwing they (rationalists) actually are.
When I first encountered people trying to expand upon the Mark Fisher interpretation of hauntology, a few years ago, I thought they were perhaps overreaching. But after a few more years of observation, now I no longer think so. The rationalist project isn’t really about predicting the future, and perhaps never really has been. Its conservative essence is to enshrine and protect a certain, mainly pop-cultural, framework about what the future should look like, and events are now passing that framework by. It only gets stranger from here.
Its conservative essence is to enshrine and protect a certain, mainly pop-cultural, framework about what the future should look like
This is such a great way of looking at it, thanks.
Of course, a lot of the creators of that future were themselves politically right-wing or libertarian. Think Heinlein.
*wishcasted
deleted by creator
If only the media would report it as “Altman is yapping some shit again” instead of breathlessly repeating his words like he’s a holy prophet
I’m sure that treating people like a dog that was tricked by wrapping medicine in a bit of bologna will go over well.
Death comes for XSLT(article about google killing XSLT, or gunning for it, not my usual fair but it is bad, and I didn’t see much attention pointed at it so far).
XSLT is basically used for one thing - styline RSS feeds so they look like HTML
(yes I know the XSLT lovers point to data sources that went all in on XML in the early 2000s then never got the budget to update, but in the first approximation that’s the use case)
Seeing as RSS is basically as popular/well known as XML, I doubt this tantrum will move the needle in any way.
Mostly people are mad at Google, which is fine by me
Here’s the relevant WHATWG issue, and it’s not just Google. Leaving aside the merits of “don’t break the web” vs “it’s barely used”, I think largely we’ve ended up here because browser vendors depended on a library (libxml2) to which they gave ~zero time or money.
Oh and, of course, in that issue the most recent comments are from a Mozilla employee saying “I asked Claude…”.
LW and the more credulous parts of the general techsphere are going gaga over this
https://rikiyat.github.io/blog/weight-space-programming/
Is this the long foretold, mythical RSI???
RSI = recursive self-improvement, not repetitive strain injury
Took one look at the post, “this is all slop isn’t it”, and sure enough the authors couldn’t even do the traditional academic’s single-commit dump without “co-authored by: Claude”.
The “recursive self-improvement” nonsense really baffles me. Excluding those who have a financial motive to talk about it, I don’t understand why anyone would think this would ever be possible for LLMs.
The “recursive self-improvement” nonsense really baffles me.
That’s always been the magic sauce that gives the AI supernatural abilities in rationalist lore, I think they cribbed it from the early singularitarians who had a thing about how when we got our shit together intelligence studying intelligence would lead to an “intelligence explosion”, except they replaced the humanity-fuck-yeah aspect with robot fetishism.
I don’t understand why anyone would think this would ever be possible for LLMs.
Yes, as I understand it, inference requires strictly less computational capacity than training; it’s just that a training setup can be turned around and run inference sessions massively in parallel. Just because you can run inference doesn’t mean you can accomplish a training run. Unless there’s some rationalist super-computer-science I’ve been missing this whole time…
lol, I saw that going aflame among the promptfondlers nearest me, too
model auto-tune. they’re getting hype for model auto-tune. I expect we’ll little outside breathless tests of autotuned models for a few days
At least we’ve moved past
classifiers that we’ve had for 15 yearsJev
He links to comments on the orange site which are pretty predictable
Author believes we need to be more reasonable about child labour
More seriously, to save you the bother of reading it, he objects to comparisons between anti-slop sentiment and the actual Luddites, because slop haters are too… centralised? And then cites our friend SE Gyges in an update.
I clicked around the blog a bit and the guy’s definitely a slopfondler. But a reasonable sloperator, of course
Addendum: so this guy has apparently read all the literature on Luddites, yet still believes their concerns had anything to do with people being ‘displaced’ by ‘automation’. Fun reminder: ‘automation’ as a word literally did not exist until 1948.
it’s not RSI because OpenAI told me they already achieved RSI on the WSJ podcast
(I’m trying to find it again but there was also an article on here about AI having none of the traits needed for RSI)
Wait will this mean that poisoning LLMs will be even harder to detect and prevent?
so there’s a new nber econometrics paper out, covering our favourite industry
it has some bangers in it
Paid adoption is substantially more common among younger and higher-income consumers: those earning more than $200,000 per year are 1.7 times as likely to pay as those earning less than $50,000, and those younger than 55 are 1.8 times as likely as those aged 55 or older
starting strong!
This force creates a dynamic incentive to set lower prices today: lower prices attract more users, improving future model quality and raising future demand
I don’t entirely follow how setting lower prices today improves future model quality but sure I’ll go along with the ride
Despite this limited horizontal differentiation, we find some degree of vertical differentiation across providers, consistent with observed market shares. At the end of our sample, the average consumer was willing to pay a $10 monthly premium for OpenAI products relative to Anthropic products. Part of this difference reflects model quality: our estimates indicate that consumers are willing to pay more for higher-quality models. For instance, moving from GPT-4’s capability level to GPT-5.5 Pro’s raises average willingness to pay for OpenAI products by approximately $25 per month
We also use the model to examine how consumer preferences for AI services shape competition. We begin by considering substitutability across services. By the end of our sample, all pairs of AI services are substitutes on average
“do you want tea or coffee?” “yes”
OpenAI has the highest retention of all AI products we study, with 77% of people still paying after one month, 65% after three months, and 55% after six months. Anthropic and xAI are lower, but similar to each other: around 65%, 50%, and 40% after one, three, and six months,
in the words of basement jaxx, down down downdown down downdown downdown
chat is it good when your world changing ai god keeps not getting picked for anyone’s team during lunch football
I don’t entirely follow how setting lower prices today improves future model quality but sure I’ll go along with the ride
More people using it, more training data fed into it.
77% of people still paying after one month
That seems a bit low.
More people using it, more training data fed into it
yes but that’s not actually the hard and costly part. the hard and costly part is the actual training action, which runs in the hundred of millions if not more
which, y’know, when lower prices……
Well, they’ve also reached model sizes where scrapes of the entire internet aren’t sufficiently large enough to train the model, so they also need more data. But yeah, they compute cost of actually training the models are prohibitively large.
yeah they’re far past public-only sources, there’s more than a few non-public sources going into this shit too. and still it’s not enough to work and sort out the issues! almost as tho the whole thing is a bit of a silly idea
I’m glad there’s finally some financial reckoning coming, I just hope I’m wrong about how utterly fucking bad it will be
If that paper about astra being able to make sense of other models’ weights has any merit, at some point they became so desperate they actually started adding serialized open weight models to the training data.
subscriptions that are less sticky than a YMCA membership
this combined with the anthropic amended s-1 (which I very briefly dug into here, will dig more later), the numbers are all just about as bad as I’ve been thinking they’d be
