Regarding a project to translate several thousand ancient letters:
So, um... this is bad. Really bad. I looked at the letters that were translated by the AI, and the very first one I found was almost entirely hallucination.
this post was submitted on 16 Mar 2026
20 points (95.5% liked)
TechTakes
2516 readers
62 users here now
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
founded 2 years ago
MODERATORS
And here I thought "people being easy to replace with a small shell script" was a joke...
fig. 1: @self hard at work keeping awful systems up and running
from Unix World 1985. enterprise computing was so much more fun in those days
Hey, we hailed our @self!
Man, if only I had enough optimism left to aspire to that level of silliness, as opposed to be sliding further and further in to the maw of computer stupidity.
Thank you @self, love your haircut
if you use Rust enough it just grows like that
A LWer is super-impressed by the time travel fantasy Illumine Lingao (an example of Chuanyue)
https://www.lesswrong.com/posts/YiRsCfkJ2ERGpRpen/leogao-s-shortform?commentId=J4YGrY26Ezt5oMsot
Listen to this pitch:
the vast majority of the book is devoted to discussing every single technical aspect in excruciating well-researched detail. you don't simply have a paragraph about them deciding to buy guns, you get an entire chapter of different gun experts arguing back and forth about exactly which gun to buy based on maintainability, range, differences between civilian and military models, semi automatic vs fully automatic.
Apparently they're quite unaware of the extensive number of works in Russian with similar themes:
https://en.wikipedia.org/wiki/Accidental_travel#In_Russian_fiction
Look, I've read some long-ass web novels. I enjoyed Worm, A Practical Guide to Evil, and Katalepsis all start to finish. I have also spent more hours than I could count (even if I did care to) perusing excessively detailed fan wikis and reading interninal debates between nerds about minutia. I have done all of this and enjoyed myself greatly.
But the way they're describing this sounds absolutely exhausting and incredibly dull. If this isn't the result of some kind of collaborative project where the debates are between different actual people then it sounds like you're just dumping your worldbuilding notes into the page and throwing a "he said" every so often.
followup, here's a real substack interview with one of the originators of the collab novel
https://afraw.substack.com/p/first-dig-the-latrines
to be honest sounds like semi-fascist shit to me.
WTF is this garbage in the Graurdain? "Let's assume!" is a terrible premise for even an opinion column to begin with, but "let's assume Musk is right and AI could allow us all to not work" is... bananas for the Guardian to publish. Even before considering that the author's bio says he's a business owner of a technology and financial management services company.
The article's entire premise is Musk saying some random shit. Remember how Musk said that he would land a man on Mars in 10 years 13 years ago? Honestly, I am incensed that people like Musk and Trump can just say shit and many people will just accept it. I can no longer tolerate it.
Putting aside the very real human ability to screw up such a concept and turn any fair system into an unfair one, ...
He says this after mentioning UBI. He really doesn't want to confront the unfortunate fact that UBI is entirely a political issue. Whatever magical beliefs one may have about how AI can create wealth, the question of how to distribute it is a social arrangement. What exactly stops the wealthy from consolidating all that wealth for themselves? The goodness of their hearts? Or is it political pushback (and violence in the bad old days), as demonstrated in every single example we have in history?
I'd say the problem is even worse now. In previous eras, some wealthy people funded libraries and parks. Nowadays we see them donate to weirdo rationalist nonsense that is completely disconnected from reality.
No getting up early and commuting on public transit. ...
This is followed by four whole paragraphs about how the office sucks and wouldn't it be wonderful if AI got rid of all that. Guess what, we have remote work already! Remember how, during COVID, many software engineering jobs went fully remote, and it turned out that the work was perfectly doable and the workers' lives improved? But then there were so many puff pieces by managers about the wonderful environment of the office, and back to the office they went. Don't worry, when the magical AI is here, they'll change their minds.
Yes, there are "mindless, stupid, inane things" like chores that are unavoidable. There are also other mindless, stupid, inane things that are entirely avoidable but exist anyway because some people base their entire lives around number go up.
load more comments
(3 replies)
An Aella-curious blogger in SoCal has noticed something:
But what I find more interesting than broadly “weird sex” is the specific interest in BDSM, kink and particularly full-contact CNC; a relatively common fantasy in individuals, but one I’ve never seen such widespread community interest in outside the Bay Area.
Kink and power-play are practices of manufactured risk, with CNC clocking at a more intense point on the same spectrum. The idea that many of these people are devoting their 9-5s and beyond to eliminating the ultimate consequence (death), only to go home and collectively play-pretend violence (scaffolded with extensive rules and consent forms) is fascinating, and- to me- makes complete sense.
The rationalist interest in manufacturing risk is the direct byproduct of their commitment to flushing it out.
The blogger attended Aella's SlutCon. I don't know if she knows that many of our friends have problems with consent as most of us understand it (their understanding is more "if they are old enough to sign the contract, and they sign, that is on them").
[Effective Altruism] was originally applied to initiatives like raising money for mosquito netting, but now includes figures like Johnson, who has reframed his blood experiments as a product of his own generosity, set to cure humanity of its greatest ill: death itself.
People keep saying this, so it's good to have a reminder that the weirdos (derogatory) were there all along.
load more comments
(4 replies)
supremely rational gamblers want to rewrite reality by threatening a journalist, because reporting got in the way of them getting money from polymarket. all while completely unaware that they're giving him better story than the actual missile impact thing https://www.timesofisrael.com/gamblers-trying-to-win-a-bet-on-polymarket-are-vowing-to-kill-me-if-i-dont-rewrite-an-iran-missile-story/ also https://awful.systems/post/7617781
update: polymarket claims to have banned users involved, not specifying how they found them https://xcancel.com/Polymarket/status/2033635318662860916#m
Polymarket when faced with the oracle problem: "What if we threaten to shoot the oracle?"
the rationalist counterpart to rubber hose cryptanalysis
e: damn it was right there: rubber hose ontology
load more comments
(1 replies)
new odium symposium episode. we examine the foundational TERF text, janice raymond's "the transsexual empire," which turns out to be about how trans people are a big pharma conspiracy
https://www.patreon.com/posts/12-invasion-of-w-152915964
www.odiumsymposium.com for links to other platforms
It always strikes me how stupid bigotry makes you. There are so many points where she seems to make a point that she cannot accept because it would go against her conclusion. Also lol @ the "not what we're called", that stuck with me for some reason
i hope you get hazard pay for all the psychic damage you inflict on each other
Did I mention that one of your more recent eps covered some shit so odious I stress ate a pile of oreos? Keep up the good work
alt-txt
Yesterday i explained something so bleak to my therapist she asked me if we could pause for a minute so she could think about it. I'm getting close to winning therapy i can feel it in my bones.
load more comments
(2 replies)
NVidia's announced an AI filter for PC gaming, calling it "AI-Powered Breakthrough In Visual Fidelity For Games" and hyping the ever-loving shit out of it.
The results are, unsurprisingly, complete garbage, and its already getting ripped apart by the gaming press.
I mean, some of their before/after images are much more impressive than the RE one, but the general look is less like a revolution in capacity and more like someone took some time to find the right Instagram filter.
Also after taking a look at Starfield's steam page for comparison I'm pretty sure that all the "before" images were taken on lower settings for existing texture quality and lighting. Like, even in areas where the DLSS gives an improvement the original game doesn't look as bad as presented here.
Also the discourse has been ongoing since at least Skyrim's original release whether or not the increasing fidelity of game graphics was actually making games better, or just more expensive to make and play. And that was before transformer models entered the picture and started cooking the world. I'm glad nVidia got some new jerk-off material, but even if it works exactly as advertised that's all it is at this point.
I'm struck by how much contrast gets blasted into the shadows of every scene, reminiscent of the average RTX "remaster." Lighting is treated not as a tool for composing scenes and guiding attention, but as a dial to be turned toward "more gooder" wherever possible. Just make everything look like everything else; that's how you know the technology is getting Better.
load more comments
(1 replies)
Maybe it's just me but even the enhanced lighting aspect doesn't look especially good, at least where faces are concerned; shining a hard light sideways so every facial nook and cranny gets highlighted in excruciating detail looks less natural and more like the old android HDR photo filter, even before you realize it's giving some characters instagram make-overs.
load more comments
(1 replies)
The Founder of Anthropic Says He Wants to Protect Humanity From AI. Just Don't Ask How. another long article about the AI craze and in particular Anthropic. A snippet that stood out to me:
"Reviewing my interview transcripts one night, I discover I’d left my recorder running when I excused myself to use the bathroom at Anthropic. On the tape, Kyle Fish, the AI researcher, and Danielle Ghiglieri, my tattooed guide, are laughing about some visitors to their headquarters the day before, what sounds like a documentary or TV crew.
“I sit right next to Trenton,” Fish says. “I went back and told him, ‘Dude, you really did something to those guys with your sunscreen stuff yesterday.’ He thought it was hilarious.”
They’re both cracking up.
Ghiglieri says Fish, too, had convincingly come off as a “different species of human,” adding: “They were very enamored with you.”
They’re inclined to cooperate with whatever project these people proposed, she says, and make everybody a star. I hadn’t heard Trenton’s sunscreen spiel yet. Only later, over lunch, would he tell me that he stopped protecting himself against skin cancer because AI was going to end the world in five years.
Crazy to me how people can so confidently predict AI doomsday, and then just keep working at an AI company
I'm more concerned that the writer could listen to this, presumably multiple times on his tape, and still wrote the rest of the piece like these guys are acting in good faith. Regardless of the unanswerable question of whether they believe their own hype, they are clearly saying things for a purpose of self-enrichment and self-aggrandizement rather than out of any concern for other people, and that is where the story should be. Even the guys most ostensibly interested in protecting humanity are still, when they think the mic is off and the journalist is out of the room, joking about how they're manipulating the press into saying what they want.
load more comments
(4 replies)
5 Tools You Can Vibe Code For Your Business In Under An Hour exactly the sort of slop from someone with a hard-on for AI, no understanding of the risks of vibe coding core parts of your business' infrastructure and guest writes for Forbes would produce.
Starts with a sickening intro that leans into "pilled" to be "down with the kids"
If you haven't joined the Claudepilled crowd, open an account and play.
Bright ideas include "copy and paste the source code from your home page into Claude" but overlooks the how to actually get those changes deployed part.
Wanna see my cool website. It's at http://localhost:1234/ take that web developers!
Then she describes building a custom internal dashboard...
Open Claude Code and describe your business. List every software tool you use. Ask it to suggest the key metrics you'd want to see from each one. Go back and forth until the list feels right. Then give it your brand guidelines and ask it to build a dashboard that displays everything. Ask for it to be password protected.
Yes that sounds like a great idea and not a car crash waiting to happen
She also describes building a customer facing onboarding site
Build a custom client-facing dashboard instead. Tell Claude Code what your onboarding process looks like step by step. Describe what information you need to collect and what your clients need to access. Ask it to build a secure portal they can log into, with automations that send them what they need and follow up to collect what you need. This is a branded, professional experience that scales without you. The emotional design matters here too: you want clients to feel held, not herded. Tell Claude that.
Yes vibe coded customer facing tools are a fantastic idea and definitely not a vector for cyber attacks nuh-uh. I'm sure it will be fine if you ask for it to be "secure" right?
FML are we in the twilight zone here?
Ask for it to be password protected.
I think I'm having a stroke. Or at least I hope I'm having a stroke and that this unparodiably dumb piece isn't any more real than it sounds.
guest writes for Forbes would produce.
I seriously think we can completely dismiss Forbes as a credible source at this point, even if it's not something coming from, ahem, "contributors"
The software industry is experiencing a huge collective AI psychosis.
New AI legal filing sanctions just dropped: https://storage.courtlistener.com/recap/gov.uscourts.ca6.152857/gov.uscourts.ca6.152857.50.2.pdf
I don't have time to read over it completely yet, but here's a taste:
That briefing repeatedly misrepresented the record, cited non-existent cases, and cited cases for propositions of law that they did not even discuss, much less support. As explained below, Irion’s and Egli’s misconduct warrants the sanctions laid out in Section II.C.
If we included typos and other errors that are arguably, but not clearly, a misrepresentation or fake citation, we would be looking at far more misstatements of fact and law
Irion and Egli did not respond to these directives. Instead, they said the show cause order was “void on its face for failing to include a signature of an Article III judge,” was “motivated by harassment of the Respondent attorneys,” and “reflect[ed] illegal ex-parte [sic] communications within this Court.”
Although citing fake cases violates Federal Rule of Appellate Procedure 38, Rule 38 alone is not “up to the task” of sanctioning this conduct, Chambers, 501 U.S. at 50, because Rule 38 allows only for the imposition of costs and attorneys’ fees, Sanctions § 33. But we think other sanctions are also appropriate, so we employ our inherent authority
Not a lawyer, just a bit of a law nerd, by this is a big deal, especially the fact that courts have been repeatedly using their inherent authority sanction on people who fuck this up. Courts do not routinely invoke their inherent authority like this. Also this footnote is interesting:
Ghostwriting is when one person writes the document while another person takes credit for it without acknowledging the true author’s identity. See The American Heritage Dictionary of the English Language 741 (4th ed. 2000). Legal authorities generally discuss ghostwriting for a pro se litigant, see, e.g., Duran v. Carris, 238 F.3d 1268, 1272 (10th Cir. 2001), but we see no reason why rules regulating ghostwriting should apply in only the pro se context. The primary concern with ghostwriting is that the true author would escape liability for his conduct, see In re Mungo, 305 B.R. 762, 768 (Bankr. D.S.C. 2003); Ellis v. Maine, 448 F.2d 1325, 1328 (1st Cir. 1971), and that concern is just as acute when a lawyer ultimately signs the ghostwritten pleading.
It sounds like they're looking for an angle to hold the LLM operators (OpenAI/Anthropic - or at least whatever company wraps the models in the necessary bits and bobs to make it a product they can sell to stupid asshole lawyers) as ultimately accountable for these filings, just as if they were a SovCit guru providing materials for one of their griftees to submit to the court without ever actually putting their name to the record where the might face consequences. I'd need to do some research to speculate on what that might mean, but it should give everyone operating in this space pause.
I'm still reading the appendix that goes into the specific hallucinations but it sounds like they're pretty absurd based on the tone of this order.
load more comments
(4 replies)
The AI-as-excuse layoffs have come for Meta:
Back and forth a few years ago on the SlateStarCodex subredit, roughly:
Scott Alexander: Bay Area rationality is wonderful, we have foundations and group homes and jolly social activities and a Solistice ritual and even "Reciprocity and Propinquity: two different rationalist dating/matchmaking services"
Rando:
I don't know, I live in a nice community in a different city where people I know have lots of Shabbat dinners, choirs, board game nights, discussions, etc. And zero people I know have joined a cult, and one person I know has developed psychosis, but she had a family history of psychosis, starting having symptoms in early adulthood, and pretty quickly went on antipsychotics and got a lot better.
Is it just that California attracts weird shit and if you put people in California, whatever they're already doing will get culty?
Alexander: base rates! how do your demographics compare to ours?
Rando:
Probably similar size and age? Nearly everyone I knew has parents who are teachers/lawyers/doctors/therapists/etc, so I guess upper middle class according to that book you wrote about a while ago.
It's not like everyone's doing great, lots of people have depression and anxiety and probably smoke more weed than is good for them. Most of those people already had those problems from their adolescence.
But our rates of weird problems, like multiple people with overlapping psychoses tied to some guy, are low.
load more comments
(4 replies)
load more comments
(7 replies)
@dgerard hey, I saw your bsky post and had an idea. Have you weighed in on Nscale and the UK's sovereign scafolding reserve? I hope that it gets noticed by the wider public, because that shit is 10/10 hilarious.
It gets better! According to Trashfuture, Nscale never even bothered to buy the scaffolding yard, which is still in operation.
https://trashfuturepodcast.podbean.com/e/unlocked-scaffold-to-heaven/
load more comments
(3 replies)
heh. thiel's sermons about antichrist annoyed vatican enough for its a.i. adviser to rhetorically ask “should we burn peter thiel?” in an article titled “american heresy: should we burn peter thiel”. as the children twenty years ago were saying: lol. lmao.
sources:
He almost certainly got the info in other places, but I find it profoundly amusing to think that in the past the AI Advisor to the Pope, may have stumbled into our corner of the internet.
I’m like 60% sure he will self-incinerate when exposed to sunlight
Dans cette vision, la démocratie entendue comme autogouvernement de citoyens égaux est déjà morte — et il ne reste plus que, dans l’obscurité d’un data center, la gestion clinique de son cadavre.
I'll grant the Holy Inquisition guy something: he really knows how to turn a phrase. the English summaries of this piece really don't do it justice.
For the non-French speakers among us:
In this vision for the world, democracy understood as the self-governance of equal citizens is already dead — and there only remains shrouded in the darkness of a data center, the clinical administration of its corpse.
load more comments
(1 replies)
Can not be rid of this rancid sub-class of plutocrats quickly enough. “Gen ⍺ hanging the last squillionaire-tech-bro by the entrails of the last kleptocrat” should be this millenium’s Diderotian coda on their existence.
Sam Altman wants his eye scanning crypto bullshit to be used to verify AI agents so he can save the internet from himself.
Rather than blocking automated traffic outright as a safety or data-protection measure, World [previously world coin] suggests sites could instead require AI agents to present an associated World ID token to prove they represent an actual human who’s behind any request. In this way, the site could allow agents to access limited resources like restaurant reservations, ticket purchase opportunities, free trials, or even bandwidth without worrying about a single user flooding the process with thousands of anonymous bots. The same idea could apply to sensitive reputational systems like online forums and polls, where it’s important to prevent automated astroturfing or dogpiling.
load more comments
(15 replies)
My heart goes out to my fleeting online acquaintance who’s seemingly but reliably two years too late to hype a trend. 2024 it was blockchain/cryptocurrencies that he tried pushing, now he’s saying AI technology companies are here to stay.
Somebody’s gotta buy the reverse reverse Cramer index I guess.
view more: next ›