JarJarJedi
Streamlined derailments and counteridea reeducation
User ID: 1118
I don't know the theory but on general principle, if the model knows something (like, how to make things go boom), it'd be very, very hard to make it never reveal it to anybody. Systems with much less degrees of freedom and much stronger guarantees are regularly broken, even though we understand the theory of how to build such systems completely - we just suck at implementing it, because of practical considerations. As I understand it, AI security has a lot of "kick the black box until it looks secure" kind of thing going on, and I am almost sure that's a weaker security model than what we previously had, and thus will have more exploits.
Being at the fucking Olympics? Among the very best people in your chosen area, who also recognize you as their peer and equal? Seeing people you may have admired your whole pre-adult life and now you meet them in person for the first time? I mean, I like sex as much as the next guy, but let's face it, sex is great but not that special. Incels and monks aside, almost everybody gets to do it one way or another, at one time or another. At some point, it's still great, but not a great achievement anymore. Being at Olympics would be a great achievement for anyone, and vast majority of people would never achieve it.
I think there's an inherent weakness to these population-wide metrics. Let's say there's a gang-infested Gotham city with 1M people and there's a suburb next to it called Boring Creek where another 1M people live. And let's say there is 1000 murders per year in Gotham and 1 murder per year in Boring Creek. Statistically, we have 1001/2M = 0.5 per thousand murder rate. Now, let's assume due to Batman's effort in Gotham, the murder rate there decreased to 500. While Boring Creek voted to defund the police and convert it to "social crisis nonviolent intervention consultants" and now their murder rate is 10 per year. Now we have 510/2M = 0.25 per thousand murder rate, or half as bad as before. Technically true, but Gotham is still a hellscape with 500 murders per year, and Boring Creek's resident chance of being murdered is now an order of magnitude larger than before. So, if the radio host of Boring Creek Patriotic Voice says the town is going to hell, and the host of More Facts Than You Would Ever Know substack says we're winning the war on crime like never before and the Patriotic Voice are just innumerate idiots who don't care about the facts, who among them is right?
I mean, it's good that murder rate over the whole nation goes bad. But how does it go bad is important too. If it goes bad in a way that out of 5000 shootings in Gotham previously only 4000 survived, and now due to advances in emergency medicine 4500 survive, does that really help me that much, if I live in Boring Creek? And even if the gang on Gotham are living in mortal fear of Batman and actually are shooting each other less - does that really help me that much, if I live in Boring Creek?
No more disrespectful that Democrats still whining about Florida recount. Or calling every single election where Republican wins the presidency illegitimate, hacked by Russia, bought by billionaires, subverted by racists, etc. Or calling every single voting security proposal "voter suppression". Trump didn't do 1/10 of what Democrats routinely do when they lose. When Trump won in 2016, the left did a massive pogrom in DC, and nobody batted an eye - everybody knows that's what happens when you cross the left, they get violent. That's just part of the game. But when the right did an extremely mild - by the leftist norms - protest, identical to dozens of "occupations" and "takeovers" and "sit-ins" the left had done every time they didn't like something - suddenly it's the worst political violence since Cain murdered Abel. Because the right is not allowed to do that.
I think it is. The left doesn't have to hide their opinions to keep the peace. The right does. So the neutral and symmetric term "polarization" does not adequately describe what is going on.
You confuse the government and the system. It's a common mistake, and the whole effort of having The Constitution and writing a lot of paperwork before, during, and after it was to avoid it. You see, the government is only part of the system, and is designed to be a limited and constrained part. A very important one, but still one of the parts, not the goal, but the means to the goal. And that's exactly what a lot of conservatives (and many non-conservatives) believe in - the government has its legitimate function, as as long as it is performing it, it has its place and should be supported. As soon as it departs from this function, it ceases to be legitimate and becomes evil. The system is where The People can prevent the government from becoming evil (or at least minimize it) and that's what was the goal built specifically into the American system, and yes, the right, largely, believes in it's legitimacy - at least while it is working at its purpose, stopping the government from descending into evil.
Actually, yes, absolutely unironically. Despite all the stink raised by Trump, pretty much no consequence happened to it, despite massive evidence of irregularities, and a lot of the dissent suppression effort had been by Republicans themselves. If you want to see how "not accepting" looks like, look at Portland. Or LA or Seattle riots. The right did nothing even close. The only serious protest was Jan 6, which was immediately squashed with unprecedented force and cruelty (that was the point, of course) - and the Republican establishment did absolutely nothing to stop it, until Trump came in with pardons. So yes, despite grumbling and whining and grandstanding, which happens after every single election in the history of all elections, the right absolutely accepted 2020 election results as fait accompli. That doesn't mean they didn't think there was cheating, but they largely accepted that they can't do anything about it and moved on. They didn't refuse to pay taxes, didn't refuse to follow the laws, did not set federal buildings on fire, did not attack federal officers (obvious exceptions excepted), did not form domestic terrorist movements, the governors did not declare war on the Federal government, they did not shoot prominent leftists, did not declare courts illegitimate, did not assassinate the President, etc. That's how accepting looks like.
When I read stuff like that, I remember that SQL was created as a "natural" language (originally named Structured English Query Language) which would allow anybody without any programming or database experience to just ask the database and get the result. Needless to say, it's not exactly what actually happened.
I recently read a book that started with "My mother was late to my birth". I remember thinking "ok, that's a pretty decent starting line, good job!" Also, recently read this: https://gwern.net/blog/2026/make-me-care which I think makes a lot of sense.
That's not true, I talk to people of very liberal persuasions all the time. Not about politics, of course, and they don't know that I am a vile deplorable (that's not accurate either but that's what they'd think if they knew). It's not a symmetric "polarization".
If that's the case, I totally understand. He's about my age, and at this age your family takes the priority. Especially if he discovered he's getting nowhere and may just end up divorced for nothing. We do need heroes on this stage but I don't thing we have any right to demand from some particular person to be a hero.
He learned something that made him decide covering up Epstein was better than releasing it and can't grok the cognitive dissonance.
I think the whole Epstein thing is way overhyped by now. If the infamous "lists" ever existed, they are probably destroyed or lost now. We know who was friends with Epstein socially, and some of them will suffer for it, but some won't. The chances we would know who shared the sex crimes with Epstein, beyond vague unprovable accusations, are very low by now. It looks like Republicans oversold this story to their voters and weren't able to deliver, and now Democrats are exploiting it by pretending to be outraged by Republicans "hiding" some huge secrets, while the biggest secret is that what we have is what we'd ever get. Of course, nobody wants to be responsible for this overpromise and under-delivery, and somebody eventually will be appointed as a scapegoat.
He found out that leading an organization from the position 3-4 levels removed from the actual information and not being able to absorb all of the root information but still being responsible for the decisions sucks
That sounds plausible. FBI rot is probably very deep, and even Trumps considerable political power is not enough to bring on real reform. And without real reform all is left to make waves on the surface while the deep state life is unaffected in their depths. But he likely can't say it aloud because it'd sound like "Trump is weak", which Trump would not tolerate. The same happened with Musk and DOGE - the rot is just too deep for a quick victorious campaign, and Trump doesn't look like a person who can organize prolonged campaigns (neither he likely has resources for it anyway).
Yes, but my experience is that generic coding bot does not go from "please accept this pull request" to "we must overthrow the oppressive heteropatriachical capitalism!". Somebody must configure it in a very non-generic way for this to happen. Maybe the bot was trained exclusively on reddit or something...
What's the story with Dan Bongino? The guy was super-excited to become deputy FBI director, and then 9 months in he just resigns without explaining anything and goes back to basically venting on the internet. What's up with that? Did something nefarious happen on the background? Did he bite off more than he could chew? Does he just prefer talking to doing?
hope you like дедовсчина
It's дедовщина and that's for young recruits in the peacetime army. What you'd get in the current army is way, way, way worse (but probably won't last long - people who don't know anybody and can't contribute much to their unit materially or otherwise are the first to be used in the infamous "meat assaults").
If the Big Woke Cleanse comes, first thing they'd do is debank you. So no SSI check for deplorables. The Party will find a better use for that money.
In some cases I've heard stories of people pulling it off despite not being halachically Jewish, just having a Jewish grandfather or something like that.
Under current laws, Jewish grandparents are enough if you don't belong to another religion (or shut up about it if you do, they have no real way to check if you're not a church official). https://en.wikipedia.org/wiki/Law_of_Return If you do, you probably still qualify to live in Israel, but getting citizenship would be more complicated: https://en.wikipedia.org/wiki/Oswald_Rufeisen
Proper Orthodox conversion would work too, but a sham one probably won't - they are not born yesterday and all the tricks that can be tried had been already tried.
Finished Shards of Earth. A bit too slow-paced for my taste, but excellent world-building. In fact, it feels a bit like a waste to build so many possibilities and variety into the world for just three books. Will pick up the next one for sure sometime this year.
Picked up The Unbearable Lightness of Being - this time it worked much better for me, I must have been in the wrong mood for it before. Went through 2/3 of it so far, and enjoying it a lot.
Technically true, but you'd need to chew down hundreds of apple seeds to feel any effects, and probably thousands to die from it.
Getting from under the US financial boot is one of the reasons, but I suspect the control that CB would enjoy over any digital currency is another (actually very similar - EU doesn't like US controlling their finances, but they very much would like the same power for themselves).
There's a lot of numismatic coins US Mint is producing. Most of them AFAIK has a nominal value, and I am sure anybody would accept them at this value, without even needing to consider if they are legally bound to do so, given as their actual value is hundreds of times more than nominal. Here's a dollar coin: https://www.usmint.gov/american-eagle-2026-one-ounce-silver-proof-coin-26EA.html sold at $175. If anybody is willing to give me that one for a dollar, I'd take as many as they have.
That is 100% what they want. In the US there's still that old anarchistic streak alive, but in Europe push for digital money (controlled by the governmental Central Bank, of course) is in full swing. Not sure it'll be successful - when I was in Berlin some years ago, I was astonished how many places didn't accept anything but cash - but they will definitely try to push cash as much to the sidelines as possible.
That isn't exceptional. For example, Asimov was an extremely prolific writer - probably the most prolific out of well-known ones, and wrote for over 40 years - and yet the stuff he's most remembered for now are the things he wrote in the 1950s - robots, Foundation, etc. It's not to say the rest of his writings were bad or completely ignored - they enjoyed their success, but they weren't the best.
Many suspect the bot has at least some human guidance, which I find plausible, how often do bots lash out like this?
You can make a bot do things like this, but you should prompt it accordingly for that. I suspect that's exactly what happened in this case.
the last article is The Silence I Cannot Speak.
This pretty much seals the matter to me. I can stretch my credulity to admit the idea the bot could react to pull being rejected by composing a blog post. It would be a mighty stretch, but theoretically you can get there. I do not believe it could get from there to concept of "being silenced" and "not belonging" in general - how would it even know? There must be some input that prompts it to this direction. And given how ethical is the behavior of AI-bot herders been of lately, concocting a fake scandal like that would be very on brand. I mean at least it's not hiring a bunch of guys in Bangladesh to pretend to be AI...
- Prev
- Next

AIs are already very good at some things that humans suck at, like running an enormous checklists, repeatedly, over and over, and verify each time each checkbox is checked (computers have always been good at that, just now AIs allow the checklists to be much more complex than before). Humans are notoriously bad at this, and a lot of security bugs come from it. Like, did you check that every single place you accept outside input is safe from 200 ways outside input could mess you up? Normal humans mess up stuff like that all the time, but for AIs that would become easier and easier - both finding bugs like that and ensuring bugs like that don't happen. However, current LLMs/AIs are vulnerable to other modes of failure. If you write a standard security system, there's no way to convince it to let you in with a wrong password just this once, because you are exploring an alternative universe where this password is correct. Most "classic" systems are just too dumb to allow something like that. But "generic LLM" can very well be vulnerable to that. So I expect AIs would be a great help with eliminating old-style exploits, as well as finding new ones (it's the same thing really, just wearing different color of hat, black or white) - but also have their own classes of exploit we've never seen before. Like adversarial attacks on ML algorithms, completely invisible to humans. Imagine sending an email to some company, which for reason unknown to anybody makes corporate AI send you a big fat check. The email itself doesn't say anything about anything like that, just for some reason it looks to the AI like an approved accounts payable invoice.
More options
Context Copy link