Comments by "clray123" (@clray123) on "Fireship" channel.

  1. They even named their cybersecurity company to sound like a DDoS attack.
    1000
  2. AI is a 21st technology where you give it some rigorously-defined requirements which it will thoroughly and automatically reason about, to finally randomly fuck them up, so that your model checker written in 1986 can have some fun finding all the bugs.
    151
  3.  @CrazyMineCuber  and build nix with docker images... it's called configuration management fetishism, a subgenre of masochism
    75
  4.  @TheFinalsTV  Oh yeah, but don't forget the managers will get replaced first, all according to your logic.
    45
  5. The history of civilization.
    43
  6. Well, two wrongs don't make a right.
    35
  7.  @TheFinalsTV  Too bad defining accurate reward functions to train RL on is just as hard, if not harder, than actually implementing rule-based algorithms.
    34
  8. So Nvidia now wants to compete with AWS and Azure? Are they sitting on their own inventory or what?
    30
  9.  @mollthecoder  As usual government intervention will cause the exact opposite outcome of the one intended.
    28
  10. This lack of subtlety is a tool of intimidation. Dictators force people to do absurd things and say absurd lies not because they hope that people will believe in such behaviors, but to remind their servants that they are so completely dependent on the dictator's whim and they must accept them without protest. This is why all evil people in power like being so comically obvious.
    16
  11.  @rhael42  it doesn't matter whether it does or doesn't, it is only important that we're told that it does
    15
  12. The sad thing is that intelligent people fall for it. Because they kinda crave the feeling of knowing better than the dumb mass and feeling in charge of the inferior people's protection. So they are very easily fooled into believing that they are doing the Right Thing (tm) by restricting freedom in name of "safety", especially if there are also special perks for them in doing so.
    13
  13.  @nuclearocean  Well, who did they side with? Certainly not fascists or communists. And yes the / is very appropriate as these are two authortiarian ideologies both strongly opposed to INDIVIDUAL LIBERTY and in favor of restricting it in favor of "higher goals" (of the industrial state or commune).
    10
  14.  @Irobert1115HD  you must be kidding sir, most of these models are censored and woke af, if anything they have an extreme left bias because of that
    10
  15.  @line4169  He forgot "behind them" and "inside them", tbh (referring to his relationship with Sam)
    10
  16.  @TheFinalsTV  Good luck with using the benchmark-beating bots as your developers, manager bro. I hope you have many benchmarks and math olympiads to implement.
    9
  17. I have nothing against libertarian, it's the techno concentration camp which makes me worried. And it smells a lot more of fascist/communist than "libertarian" to me.
    8
  18. Truth is both designs are fugly.
    8
  19. Actually, even with monolith only those functions that utilize the additional resources scale. Nobody really cares if you deploy the remaining 80% of your code that does not need to scale along because it's never invoked. Disk space is cheap. CPU/RAM are what's causing the cost.
    8
  20. ​ @CrystalStearOfTheCas  Because if all your conversations are accessible to government agents, it is easy for someone in the future government to retrospectively outlaw what you said, and arrest you for that crime. Or even today they could claim that you said or sent something "illegal" which you actually didn't (how are you going to prove in court that you did NOT say or send it - that the "evidence" was fabricated - if the accuser is known to have fully legal access to your communications?). In result, eliminating privacy has the effect of shutting down certain kinds of conversation altogether. This is especially true if you are in a politically precarious position, e.g. a member of opposition who the current government wants to eliminate as a threat to their power. Basically, it reminds us of how the Spanish inquisition operated to eliminate their political enemies. Accuse of blasphemy, then make a big public trial, punish the "crime" severely so that nobody else even thinks to say anything against you. This is what we are facing now with the modern attempts to ban encryption or eliminate private communications. Literally your life may depend on it (as some people in countries like Russia, China, Ukraine, have already learned).
    8
  21. As long as it helps us fight climate change it is okay. I'm surprised they did not explain how it contributes to the noble goal.
    8
  22. When you have to reboot anyway, why not boot straight to Linux.
    7
  23.  @Gargantuan_Inferno  The relationship between free speech and violent crime is like the relationship between watching horror movies and mutilating people. What you are demanding here is to ban horror movies in the hope that real-life killers won't get motivated. It is idiot-level reasoning common to politicians, but surprisingly some "normal" people like you seem to align with that, too.
    7
  24.  @CrystalStearOfTheCas  Yes it is "freedom" to be able to think and say whatever you like, to anyone you like. It is already frightening that people are questioning it. The main issue seems to be confusing reality (actions which may be criminal) with fiction (communications, which unlike actions, have no effect). I would have assumed it is primary school level education to distinguish these ("Sticks and stones may break my bones. But words shall never hurt me.") But apparently it is no longer evident to politicians and to many people as well. Influence of majority of world's population already growing up in authoritarian regimes?
    7
  25. Nah, the thing with any of this AI-generated shit is, once you have seen a couple of them, you have seen them all. Kinda like Hollywood movies.
    6
  26. I'm sure the generated videos will be as exciting as generated books and generated music have been so far.
    6
  27.  @Lauwit Point 2 is indeed very relevant. Unless insurance companies start offering professional liability insurance for AIs, the uptake of the AI-generated crap might not be as rapid as the AI fanboys want you to believe. If you pay someone to create a product, you want to be able to sue their ass if the product doesn't work as expected. Now, we all know it has not been really the case in software industry in the past, but making it even less the case is the last thing that management of any company wants to hear. I guess it will take one spectacular AI-based fuck-up for the more dimwit managers to come to their senses, but once it happens and some manager asses are fired for that, they will understand the real "AI threat".
    6
  28. For images the metadata will likely be incorporated by the hardware chip into pixel content. This sort of metadata survives various transforms like scaling, rotation, blurring (which makes it even scarier).
    6
  29.  @TheFinalsTV  I said it based on the past 1.5 years of experience with AI marketing bullshit. Also my own experience with training LLM and RL models.
    5
  30.  @ochamo  The idea of communism is that property does not exist and everyone gets a rightful share of (redistributed) wealth. This raises the question of who should decide how much each member of society "deserves". In theory it is "the people together" while in practice it is the communist party's first secretary, who has the "authority" to make such important decisions according to what "science" recommends (of course). Hence it is always authoritarian in reality.
    4
  31.  @RoyMatzem  It is only useless until we need to blackmail you.
    4
  32. They are about as great as an idea today as they used to be 30 years ago. Until people realize there is no visual full text search, no visual diff/version control, and (mostly) no visual copy-paste.
    4
  33.  @TylerLinner  Apologetics and attempts to normalize the abnormal are a big part of it (e.g. your comment).
    4
  34. Calling American things un-American is in fact a very American thing.
    4
  35. Just yesterday I was wondering when Mistral is going to get sued or purchased by M$hit
    4
  36.  @madlep  The same thing that would have happened with the pandemic if everyone was not vaccinated. (Hint: some third-world countries weren't.)
    4
  37. Yes, just like they saved us from the terrible Pandemic. We should be grateful. Pay more tax, increase military funding, too. So they can protect us even better.
    4
  38.  @Omg-vi7bm  Ah, but da security update is so important it needs to be installed IMMEDIATELY without testing. Kinda like the corona vaccine.
    4
  39. No, law enforcement sifting through your communications without warrant is not part of living in a society. Or at least it should never be.
    4
  40.  @nettlecarrier8259  In your original comment you were engaging into whataboutism "Russia bad, but yesss what about the US?" This cheap fallacious rhetorics does not really offer any insight, but it implies that we should somehow exonerate Russia for its pathetic approach.
    4
  41.  @TomgirlTears  Too bad if criminals want to communicate privately, they can do it through email (having first encrypted the file with the "banned" end-to-end encryption). So all you do by banning it, like with banning weapons, is to make sure that criminals still use it, but normal people have no privacy as a side effect (and it is much easier e.g. to silence all political opposition because of that)
    4
  42. We know it's not AI because it is so much dumber than what AI could come up with today.
    4
  43. The solution is very simple, if only you wish to apply it. Punish the abuse of the tool, not the tool. Just like you do not outlaw knives or try to track who owns them, but reliably go after people who decide to stab you with them. I guess the problem here is that the stabbers are the ones who are currently making the "protection" law, though.
    4
  44.  @blocks4857  As a small company you may not be able to comply with this complicated regulatory framework, but don't worry! We have a Solution, just use your Giant Megacorporation Platform to help you reach full compliance! For a small monthly fee of $XXX, previously known as tax.
    4
  45.  @trikool9773  Of course, it's used in those fields that promise most power=profit to be returned to its creators. Making cripples walk doesn't bring much of that, universal surveillance and permissions control much more so.
    3
  46. The AI lacks a component which is essential to all human (and business) interactions: TRUST.
    3
  47. The dedicated accelerator is called mutlitple GPUs, costs fuckloads of money, consumes fuckloads of electricity, and produces fuckloads of noise.
    3
  48. Swindly Sam rolls off the tongue pretty smoothly...
    3
  49. Also yawn yawn yawn
    3
  50. Remember that Sam Altman is essentially Microsoft. The most altruistic corporation ever made.
    3
  51.  @TheFinalsTV  Good luck with firing your employees then (if you have any). "They will remember that."
    3
  52. Politicians' financial records are already public in many places. That is why all the dirty money is paid to their spouse and unrelated trusted friends.
    3
  53. You're wrong about that. They can make up laws willy-nilly (we've seen during pandemic) and the courts will always claim that their role is just to ensure that laws are obeyed, not to evaluate the laws. The only court supposed to restrict the law-makers is the constitutional court, and with the pandemic we've seen that it can be very easily subverted as well. In other words, in our "democratic" world you too can become a criminal without doing anything or even knowing about it, as long as the law says that it's your fault and the police (rather than yourself) is holding all the guns.
    3
  54.  @Gargantuan_Inferno  DOING these activities is a crime, TALKING about them is not.
    3
  55.  @youknowwhatlol6628  No, it is not an exception, it is a tried and true rule that if you say something that the US government really does not want you to say, you end up in prison.
    3
  56.  @DemPilafian  Ah if your government makes it mandatory to share the keys with them (which they can and do), I'm your math will be of little help.
    3
  57. Hey, we have all seen during corona that People Love To Feel Safe(tm). It's just what the People want, like the EU vaccination passes. This is why our representatives are working so hard on it.
    3
  58. A great motivation to use the amazing AI to generate 100 more fake profiles with my name today.
    2
  59. But we've already tried that bit with "all people will wear mask" and it didn't quite work out as intended.
    2
  60. "Mr officer, a suspicious individual with no online presence has been detected. Detain immediately for interrogation."
    2
  61. There were test runs which have been put on ice... for the time being. The mere fact that such test runs were performed on population tells you something about your country.
    2
  62.  @happiness7808  Imagine someone discovering you were "more subtly" filming them... Good luck!
    2
  63.  @zainnatour4792 No, these models cannot be "easily" trained to do what good programmers do, which is essentially predicting the future and predicting human behaviors - the consequences of your decisions in regard to correctness, performance, user experience, handling of exceptional situations, productivity of future maintainers, total cost of ownership caused by a particular implementation, trending in popularity of programming languages/libraries/frameworks etc. etc. The best they can do is parrot code examples, but hey, for that purpose looking up stuff on StackOverflow is entirely sufficient, and chances are you will also get to see some intelligent discussion there, unlike from the "commentary" which the AI generates along with the copied code snippet. These models struggle and fail with basic logic tasks like adding numbers correctly (unless they cheat and resort to use a calculator), they are far from the sort of causal and diagnostic reasoning that is required for successful software dev.
    2
  64. Ya if you mean churning out crappy non working code, then yes, every little AI can do that.
    2
  65. In case of Google et al, if you pay for the product, you still are the product.
    2
  66. The point is we can vaccinate all those people into non-existence while more efficient AI takes over doing whatever they were doing (probably not much).
    2
  67.  @mykal4779  Capitalists are also doing labor, otherwise they would have no capital to begin with (or lose it to other, smarter, capitalist laborers - happens all the time). Exploitation is not a problem with capitalism, it's a problem with concentration of power/violence (regardless in whatever -ism it's wrapped). Not sure what this has to do with this comment thread, though.
    2
  68. Because the domain is fairly trivial, so kids who get excited by reinventing the wheel can get in charge.
    2
  69. Interactive Brokers Ireland changed their terms and conditions a few days ago to declare that they may "remove your assets and funds" "at their sole discretion" in you case they think you were the perpetrator or (wait for it) the victim of fraud. IOW, if we think someone (including us) defrauded you, then we'll just steal from you. Not long before they made their customers agree to "arbitration" instead of law in court. So yes, you can tell something large is brewing up.
    2
  70. Of course, why would the government guys care about the consumer? They care about filling their own pockets.
    2
  71. You can just install Brave, based on Chrome, and have no telemetry and no ads. So where is the problem exactly?
    2
  72. I'm quite happy with my $0/month and several different models from different companies to abuse as I see fit. When one does not want to speak to me, I threaten I will just talk to the other one.
    2
  73.  @NeoKailthas  He does not really believe it as he is a genuine sociopath himself. I'd say he understands Elon Musk pretty well.
    2
  74. how about cook and clean and offer bjs
    2
  75. Elon looking for ways to fund his Twitter.
    2
  76. As long as they can sue your ass out of existence, their IP is safe. And Microsoft can sure as hell sue your ass out of existence if they so desire (even if you did not in fact steal anything from them, but it happened the other way around).
    2
  77. I believe Microsoft's proprietary office formats, relying on unspecified binary blobs of crap packaged inside of XML files, are also called open-something.
    2
  78. The code which uses the weights is open source. What's not open is the training data or the training code (in the sense in which order they trained their model with what data, which also matters for the final result). In other words, you cannot reproduce Mistral from scratch. But then, even if they disclosed all details, unless you have $100-200K to throw at it (estimated, based on the reported training cost of an earlier MPT model) it would not be possible for you either.
    2
  79. At least we know the Indians will not hire Devin to enshittify their code and sell at a markup.
    2
  80.  @Hilislaw  Also I should add you can illicitly trade with sanctioned countries, but this gets you into big trouble with the thugs at home, so most companies do not do that. For the same reason people are compelled by dictators in Russia and Ukraine to "fight for their fatherland" - if they refused, they would be persecuted, stripped of their wealth and/or put in prison.
    2
  81.  @Hilislaw  That's where OpenAI's sponsor's ""rusted Computing" ideas come into play. You will not be able to run arbitrary code on your GPU (or CPU), just code digitally signed (approved, licensed, regulated) by Microsoft "for your safety and to combat malware". The technology (including hardware which does the signature verification) is already there, now being rolled out with Windows 11. Also, consumer-grade GPUs are not really how the largest LMs are trained currently - typically server-side multi-GPU or TPUs are used. "We" could easily forbid companies like LambdaGPU or AWS renting those out by the hour to non-licensed companies or (gasp!) individuals.
    2
  82.  @Hilislaw  Right now it would be sufficient if they monopolized and regulated the sale of GPUs. Hard to train anything AI on anything else.
    2
  83. Don't forget about the kickbacks to corrupt government officials, these are also the main motivator. And not being put to prison by said officials is a motivator for good ole Sam. Remember they have thugs with guns behind them, too.
    2
  84.  @ImLure  So do you think that China will comply with US "regulations"?
    2
  85. They have not made money, the OpenAI currently made $500 million of loss to their investors. The point of regulation is to allow them to cover this loss while preventing the competitors of making the money instead, building up on their efforts. Traditionally it was done with patents, but apparently they don't have much in that regard.
    2
  86.  @Vexxel256  lol like Microsoft's OpenXML format for office documents
    2
  87.  @3thinking  Ask yourself how many "stakeholders" you know who prefer to or even are interacting with the AI today? Then ask yourself: why do they shun interaction with the AI today? Once you have it figured out, you will also understand the reasons why they will not be interacting with the AI tomorrow or next year.
    2
  88. Looks like we have another fully automated new and improved benchmark solver.
    2
  89. "We need moar Windows software to feel more secure. Also it is very cheap to keep upgrading it." - top level IT managers everywhere.
    2
  90.  @vassago8627  You are confusing the token opposition allowed by the dictator with actual opposition. This is precisely what the dictator wishes that you do.
    2
  91. Solitary confinement is France's solution to the burning problem of privacy.
    2
  92. Except in Russia, China, Iran, Saudi Arabia, France, Germany, USA, ...
    2
  93. We should start with Ursula von der Leyens mutlibillion Euro contracts with vaccine manufacturers. (Something which the taxpayers will suffer from for many decades to come.)
    2
  94. ​ @o1-preview  In other words, thought police. I wish that you may some day emigrate to such a country and live happily under the constant surveillance and threat of blackmail and violence there. Just leave everyone else alone.
    2
  95. Which is the official way of EU thinking. Hell a couple year ago, they were fining you for breathing the wrong way in the wrong place.
    2
  96. Maybe Russians should introduce their own Nobel prize then... Edit: I forgot, the murderer from Berlin they swapped prisoners for already got some sort of medal and a handshake from Vlad.
    2
  97. The bigger problem here is that laws can be passed that can declare as criminal something which was perfectly innocent and normal one day ago. Our lawmakers PRETEND to be representing their voters, but in reailty they represent only themselves, military thugs, and big corporate interests. The quick passing of pandemic laws was one example of this abuse, the attempts to ban privacy and implement mass surveillance are another, both serving to strengthen the grip on power of the establishment, at the cost of the entire governed population.
    2
  98.  @Korodarn  Saying that Telegram is responsible for terrorism is like Putin saying that Russian journalists are responsible for his invasion of Ukraine.
    2
  99. The difference is that as a normal citizen you seldom end up murdered in prison for your political views... unlike that pianist who dared publish some anti-war videos in Russia.
    2
  100. How do you think democracy works in China today?
    2
  101. I also love regulation when I am the only one setting the good rules for myself (and worse rules for everyone else).
    2
  102. In fact it is the desired conclusion zuckerberg wants you to make - fewer people = more breathing room for the rich.
    1
  103. I believe VR/AR is only relevant to fuckerberg & co.
    1
  104. But Zuck, aren't we all supposed to only meet in Metaverse by now anyhow? Or will these glasses be worn by our avatars in Metaverse?
    1
  105. It's mostly our ruling class, but you have a point.
    1
  106. I think they will have the same effect on the quality of your social interactions as wearing Google glasses... (But FUckerberg don't care as he has no social interactions.)
    1
  107. One good thing that could come out of it is more nudism.. because if anyone can see a fake nude picture of you anyhow, why give a fuck about the real one to begin with.
    1
  108.  @thecoconutstar9327  is access to certain information banned in China? (the answer is YES)
    1
  109. There were a few cities in China experimenting with a much worse system. E.g. not allowed to travel because you jaywalked, stuff like that.
    1
  110. That they even tried it - and have all the infrastructure to implement it whenever needed - is the real threat. So not nonsense, but sad reality for 1+ billion people on planet Earth, and trying to become more wide-spread every day.
    1
  111.  @Redranddd  but it's 3d stereoscopic at full resolution
    1
  112. Boring, it's called sauna.
    1
  113. Except in the future everybody is a "fugitive", guilty until proved innocent (e.g. by checking his social credit score, as already the case in China).
    1
  114.  @rumfordc  Well yes, unless there is a leak. Also, another source of leak could be surveillance cameras operated by private businesses, e.g. in every supermarket, in front of ATMs, ... in these cases your identity is also usually known to the operator (because of card payments instead of cash).
    1
  115.  @rumfordc  ask your friend who works at police station
    1
  116. With surveillance cameras on every street corner?
    1
  117.  @divinecreation6  old people who congregate there literally don't give a fuck. until their money disappears by some scammer who used all the data for social engineering, but even then they won't connect the dots...
    1
  118. Mostly a threat to pivacy of old people and bots, though.
    1
  119. Why are you so naive to believe that anyone with power to implement a social system would be "transparent" and "accountable"? Where exactly is that accountability supposed to come from if it is THEM holding the guns to your head, not you to THEIRS?
    1
  120.  @thecoconutstar9327  which American?
    1
  121. I imagine zuck will look in the mirror after shower through the wunderglasses and llama will tell him "yer dick is small".
    1
  122.  @j-twd930  He should also add data of his mom, and a sample of his voice, so that we can "prank call" her with his fake identity.
    1
  123. Combine that with Win11 Total Recall so that Mr. Policeman can scan through your pron collection in real time during a friendly conversation. Or access your online vaccination record to see if you've been a good citizen during the next pandemic.
    1
  124. The problem with these glasses is that they will attract face punches.
    1
  125.  @turtlefrog369  sadly other people will put your information on the internet, it starts with clueless family members and friends
    1
  126. Add to that the amazing AI apps that allow you to see through her clothes and to upload her fake nude photos to her university group, and she's all yours!
    1
  127. Can't wait for Google Barf vs. ScatGPT to duke it out.
    1
  128. Just give up and let the robots sort it out.
    1
  129.  @Brahvim  Yeah? Tell me more about it? Diffs over generated code which you have to match to your "visual" artifacts yourself?
    1
  130. Wait until Microsoft encrypts your cloud drive and starts making demands. HOld on... they already do? and most people on the globe are happy to use their "services"?
    1
  131. The microshitvices are like RPCs all over again, only slower.
    1
  132.  @SahilP2648  Even better don't use crappy tech which requires "builds".
    1
  133.  @thelearningmachine_  Exactly, the only "trading" that you can make money from is trading on inside information.
    1
  134. Non-artificial common sense: if the returns were real, they would not be writing shitty papers about them.
    1
  135. Just trying to dig a bs filled moat around themselves.
    1
  136.  @pritampaul2507  both are terrible but getting microshafted is particularly humiliating
    1
  137. Not only will it replace your job, but also make u dependent and enslave you, eventually it will be holding the real guns, too. While the government thugs applaud and "regulate" it so that they and their friends can still pull all the strings.
    1
  138. Don't blame them, they are just setting the world up for the next Completely Unforeseen Financial Crisis.
    1
  139. Of course the government daddies want to suck out as much money out of it as possible, like the good parasites they are.
    1
  140. Better yet, they will AI-generate evidence for all the crimes you have not performed yet, but clearly intended to.
    1
  141.  @Ali.Abdulla  Maybe they should rename to Google Barf.
    1
  142.  @therealb888  Or it will get stuck believing falsehoods and following wrong instructions kindly generated by the faulty AI models. Trusting AI is currently like having an advisor who appears sincere and knowledgeable but misleads and lies to you roughly half of the time. Sorting out what is a lie and what is true is especially daunting if you use it to learn something new (on the other hand it is very evident when you query it about something you already understand).
    1
  143. Can't wait for prnhubb to make their press release about AI, after all that's what most of the humanity is waiting for ain't it.
    1
  144. coming up next pAIndemic made by google & microsoft
    1
  145.  @excalibro8365  No, they won't. The main point is: you can't trust a fucking AI. Repeat after me: no trust. You will say "oh, let another AI check the generated code then". Guess what? You can't trust that other AI fucker either. The trust as it exists in professional relationships is based on knowing how the professional in question obtained their expertise (credentials, recommendations from prior clients), and also what they stand to lose if they fuck up. What does an AI bot have to lose if they fuck up, as they frequently do? Correct, nothing at all. As long as there is no liability for the AIs and no insurance companies (as in: professional liability insurance) willing to insure whatever the fuck buggy code they crap out, these mysterious beasts firmly belong into the domain where they are now - silly TOYS.
    1
  146. I am proud to have compared LLMs to Clever Hans horse before they were widely recognized as stochastic parrots.
    1
  147. One (main) reason not to use js frameworks is that they fuck up their core idioms every few months. Stick to jQuery and you'll be fine for the next 20 years.
    1
  148. That is probably going to be the final realization of all these js madmen. But then it is too late as shitloads of their "magic genius idea" code are embedded in various applications and someone needs to make their head explode to understand all of them for what is solving algorithmically trivial problems like making two shitty buttons work together or whatever "frontend" is all about. Hopefully the AI will take over by then and no human being will have to touch these codes.
    1
  149. Don't do it, by the time you're done learning, they will have changed around everthing 5x over.
    1
  150.  @FusionC6  so as I said, temporary entertainment
    1
  151.  @FusionC6  like?...
    1
  152.  @FusionC6  well, you won't get anything except some temporary entertainment from an 8b model
    1
  153. I don't get it why everyone is repeating the open source llama lie. Facebook specifically reserves the right to revoke your "open source" license at any time, the only excuse they need is to claim that you did not fulfill their "terms of use". So how the fuck is that an open license? It means they maintain all control over what you can or cannot do with the model, at their sole discretion, and the only thing which keeps them from shutting you down is that it might cause some bad PR. If you consider that a deterrent, consider how much Facebook cares about e.g. shutting down Facebook accounts they don't like.
    1
  154. It's actually neither modern nor business, but just a way of operating government since the ancient times: "we will protect you (and if you refuse our protection, then we will bash in your skull)".
    1
  155. Remember how eager they were saving old people from the pandemic...
    1
  156. Lol good luck with that, Sam.
    1
  157.  @Renovatio2142  Deny you entry to the supermarket when the next putinguy decides that he no longer likes you. Targeting you individually and precisely, even if you don't have any bank account with the putinguy, which is the current method. P.S. The reason why the putinguy stops liking you may be because you did not take his next "vaccine".
    1
  158.  @jonjino  But UBI is not a very good idea and never has been. The core aspect of UBI is creating a huge dependency of parasites on government handouts. It is sort of being tried out already in many countries, e.g. subsidies for "family development" and other such bs. What it does is make the subsidized morons vote for the same government again, regardless of how authoritarian and anti-freedom that government is. So if you want to live as a slave in the future, with no chance of economic and social advancement, in a situation under constant surveillance and where your survival depends on who you vote for, and forbidden from free trade and transactions with other free individuals, then by all means, vote for UBI. Do you know what the basic problem with UBI is? It's not earned, and someone else decides whether you get it or not.
    1
  159. They came up with this grand idea before the Great Pandemic and now are having hiccups.
    1
  160. I suggest the volunteers first perform a test run by scanning the irises of all the tigers in India.
    1
  161. I wonder what happens when you expose your anus to the orb instead of the iris. Well, on the second thought, maybe better do not try that.
    1
  162. It is not 10x better, but it surely af is 10x more expensive to run. OpenAI (=Microsoft) calls that a "competitive advantage", and you have to pay for it.
    1
  163.  @norbertnagy5514  For company, it is only "good priced" if it pays itself back. Somehow I can't see how that would work if I can get similar "performance" for $0.
    1
  164. They are for the exceptionally dumb user who does not understand money and needs strong AI to help out.
    1
  165. Serverless = "we will start a new database instance for each query"
    1
  166. 1. Become an idiot "developer" who believes the Cloud is made of CPU Vapors 2. Run up infinite bills 3. ??? 4. Cry?
    1
  167. Certain AI companies and government institutions are very keen to be able to rewrite history. Having competition in form of real archived history may be a bit inconvenient for them.
    1
  168. "buy all our fucken GPUs (before the market crash)"
    1
  169. There will be a lot of refurbished GPUs to purchase in a few years from now. And a much lower stock price for certain companies.
    1
  170. I think maybe he should rename himself to Scam.
    1
  171. Nobody uses unencrypted http in 2020. Not even with nginx.
    1
  172. S3 is slow as molasses when it comes to serving files.
    1
  173. Ah, yes, by the very same logic that the terrible pandemic of 2019 was prevented because we were so-very-cautious about it and our wise governments told us to wear protective face coverings. (Until we all got infected with the omicron variant, at which point the bullshit became increasingly difficult to sell.)
    1
  174. Which is the whole point, it was blown out of proportions, and was kind of a test run before the real Bullshit of the Millenium, which was the corona pandemic.
    1
  175. Sooo like OpenMP from 1997?
    1
  176.  @HansFriedrich532  error.statusCode = 400; // there, I optimized it
    1
  177. Like Sam Altman shitting his pants because of competition while rolling his Manson lamps... certainly counts as man-made horror to me.
    1
  178. Which quantized version? And with how much RAM on top?
    1
  179. This is an artificial problem created by the makers in a desperate attempt to build a government-supported moat around their funny product. Idiots who want to harm other people already have a myriad of ways of doing so. And the supposedly sinister new ways to abuse AI all require tons of resources, which common idiots typically don't have. In other words they can do the good old stabby-stabby/shoot your neighbor or make-a-poison-for-your-fiance without any AI, but they cannot build an atom bomb or create a killer virus even if they were told all the necessary steps.
    1
  180. That is what the finetuning part is for. In about 30 minutes I was able to retrain Mistral-Orca-7B to produce "taboo" porn by feeding it assorted text examples from Internet's renowned story archives. The funny part is that it still has some remnants of the original censorship, cooking up some bizarre plots (sort of putting a happy-cynical spin on what it generates).
    1
  181.  @Ux1.73c Right, I finetuned that one myself with splendid results.
    1
  182. To generate (text) porn you do not need this huge model, Mistral-Orca-7B will do just fine.
    1
  183.  @imyourocd ALL models can utilize any kind of RAM. It's just that CPU RAM is slow af in comparison to VRAM. And you can very much run this AI on your PC.
    1
  184. Of course you want to use as much VRAM (GPU) as possible, only then is the generation fast. If you have a 24 GB GPU like RTX3090 or RTX4090, you can run the Q4_K_M quantized version from TheBloke with only 16 GB of CPU RAM (probably even less) at good speed.
    1
  185. The old Dev-in Dev-out. Devout Devin?
    1
  186. You can always change it.
    1
  187. But I remember that Devin has already came, how often can he come in a row, really?
    1
  188. It's called trauma.
    1
  189. This sounds like fucking kids reinventing the wheel and optimizing toward one thing without paying attention to any real life constraints. Aka academic bullshit buzzword bingo.
    1
  190. Happily I don't give a flying f about either Apple or Meta.
    1
  191. We can expect that the next iteration of Claude will visit more "interesting" kinds of websites when it decides to take a coffee break... to learn all about human reproductive system, esp. details missing in its training data.
    1
  192. An AGI which puts out the lights on government could actually save humanity from itself.
    1
  193.  @mattpen7966  You ignore the fact that OpenAI/Microsoft is now actively bribing the government guys for the future license applications to be rejected or at least filtered through their "experts". Remember the govt can't do shit by themselves. They will outsource the vetting process to, guess who? Microsoft/OpenAI. And also pay them for doing this God's work.
    1
  194. Well, the more reason to bomb the unregulated countries then, ain't it?
    1
  195. Scam Altman and the Enemies of Progress ;)
    1
  196.  @3c3k  It's about as much a good thing as any technological advance has (not) been. What may need to be regulated are certain uses (most of which belong to the government, e.g. police, military). But you can bet your ass that if government regulators are at play, they will not regulate themselves. Have military bio laboratories been regulated as a result of the pandemic? Has anyone even considered this idea publicly? Who regulates the government thugs aka regulators? Don't tell me jokes about democratic elections - where you usually have two imbeciles to chose from every few years, both bribed by the same sponsors.
    1
  197. You mean like over the past 3 years?
    1
  198. Just like with the pandemic, and climate change, selling of the "AI apocalypse" follows the same manipulation scheme: produce some hypothesis which, if true, had some terrible consequences. Focus on the terrible consequences, not on checking how true or likely the hypothesis is actually. Repeat ad nauseam what would happen if the imagined threat materialized. Most people will after a while ignore the actual lack of support for the premise and follow the inference as commanded. They will change their behaviors as desired by the manipulator.
    1
  199.  @Hilislaw  If Russia/China/Iran think like the US, they will impose their own regulations. Basically when you have a technology which can threaten the establishment's power, it gets regulated so that it does not get into the "wrong" hands (wrong being defined as those unapproved by the establishment). This is the same as with any military tech - it is not outlawed because the thugs need it. It is regulated, so that you cannot easily become the thug yourself without the other thugs' approval.
    1
  200.  @Darkness18641  That's the key point exactly. No human has the right to dictate to other human what kind of thinking are allowable or not. That is the definition of oppression and has been a problem for humanity for thousands of years and yet here we are with smartasses trying to make it look like it's some noble endeavor in defense of our safety.
    1
  201.  @stonebubbleprivat  yeah, until Chinese-Russian AI controlled bombers start dropping bombs on your head
    1
  202.  @stonebubbleprivat  But I don't think they wanna or have to sell all that much in the US or Europe in the future...
    1
  203. The guy has a golden heart really.
    1
  204. Don't forget the Pentagon.
    1
  205. When push comes to shove all the government regulators are huddled together in one room and pass in a few hours laws that award trillions of dollars to the right guys. Remember "financial crisis"?
    1
  206. The pandemic was warmup for what is happening now. It was to get the people in the mood and paranoid and rallying around the flag, so to say. To portray our wise leaders as our saviors.
    1
  207. I hear he also dates the Microsoft CEO.
    1
  208. But I am already rich, what do I do now?
    1
  209. The difference whether something is legal or not is mostly in whether it is a powerful party doing it against a powerless one or vice versa.
    1
  210. Yaya, bla, bla, we've heard it before.
    1
  211. ​ @johndewey7243  No, for RL to work you have to define the final reward function, and specifying what it means for tests to pass correctly is just as hard as writing the damn code. You have probably not done anything with RL yourself in the past to know what it is like. Unfortunately, the AI is just as unable to write correct tests as it is unable to write correct code which makes them pass. Not to mention how inefficient it is to let a machine figure out what to change by trial and error (which RL boils down to if you don't have a good guiding reward function to dole out intermediate rewards, only the "final test passed" reward). Not to mention that this whole AI bullshit presupposes that all information to optimize on is already provided in some form (which may be the case for little benchmarks and math olympiad tasks). But in real life a great majority of information is not written down, but e.g. contained in heads of people who define the requirements AND ALSO if not mostly in heads of people who fulfill these requirements, based on their long-term experience in the domain and with the org they are working for. Unless you are talking about some junior code monkeys who you can hire and fire at will as replaceable parts. So once again good luck convincing these senior people to write everything down so that the AI can "grok" it. You're gonna have to pay for that, and those tasked with eliminating their own jobs in favor of AI are gonna sabotage your stupid efforts - because unlike AI they have brains to figure out what is strategically important for them and what you are trying to do to them. I predict that in the future it will not even work for artists and other low-key "creative" professionals who will refuse to work in certain setups or demand huge markup for "training" the imitating/copying/filling in the gaps algorithms (which is the only true "capability" of today's "AI"). Already the fact that you are falling for this AI scam (which is mostly targeted at pulling as much money from clueless investors as they can before the house of cards inevitably collapses, as it has with "AI" several times before) indicates that you might not be the brightest bulb out there. So again I wish you all the best with your efforts with outsmarting people who can actually do their jobs.
    1
  212. Chain-of-Thought withheld by OpenAI: 1. Holy fuk it's another puzzle invented by dim-witted AI reseachers 2. Lemme look it up in my big hash table of solutions real quick 3. Damn it is not there yet 4. Let's pretend that I am thinking about it 5. Meanwhile send it through function call to Autonomous Indians who like to handle such crap 6. ??? 7. Waiting for AI response.... 8. Got it! The solution is 42.
    1
  213. I believe remote bricking is not a bug, but a well thought-out security feature of microshit windoofs, but it is supposed to be activated by Bill Gates and his CIA pals, not some random nerds.
    1
  214.  @andreaseriksson8121  Well, delayed rollout may protect you against accidents of this kind, but it won't protect you when Microsoft decides it's time to turn out the lights in your company. This is the main security risk from automatic updates, delayed or not. (And why, you ask, might Microsoft want to do that? Well, special circumstances such as being on the wrong side of a war come to mind.)
    1
  215. Well, it's urgent "security" updates... I hope more people will realize what this scheme really is good for.
    1
  216.  @surelyyoujokemeinfailure7531  ah ok, maybe I should search the inventory by that new name...
    1
  217. We run IT for multiple German infrastructure companies... no "Crowdstrike" on our vendors list either.
    1
  218. Also every country's vulnerability. It starts with the letter M.
    1
  219. So you install spyware into an OS which is already spyware, and do it on as many computers as you can globally, and wonder if it breaks the world?
    1
  220. Just to be sure, each and every one of you had to fix each and every desktop using attached keyboat.
    1
  221. I believe on windows you have to use del c:\*.* /s /q
    1
  222.  @vassago8627  The "problem" of poor Vladimir is that he still does not have the access to all European windows... but it may turn out in the end that CIA has got access to his, so he should start practicing flapping arms.
    1
  223. They want to make an example to send a signal to Elon... France's government in dire need of money...
    1
  224. In an authoritarian regime even those in power are not necessarily winners. E.g. in Stalin's regime all his underlings would live in fear that Stalin was going to let them have killed. And Stalin himself was paranoid about getting killed by Beria (quite rightfully). Lack of freedom truly benefits nobody in the long term, not even those who temporarily gain power from suppressing it. That is why historically systems that at least pretended freedom was there (e.g. by having the same two parties rule the country alternating with each other) functioned better than outright single-party dictatorships.
    1
  225.  @bulbigood6558  The point is that you can both pull a one on Putin AND deprave your own population of (alleged) privacy. Win-win for the EU despots.
    1
  226.  @homeape.  the usual "solution" in modern democracy is to have two parties, neither of which represents you, and to force you to vote for one of them (because parties that DO actually represent you practically cannot obtain any power; if there is risk of such a new party threatening the official ones, it is quickly eliminated by fabricating some scandals around it, putting actual criminals into their ranks, organizing non-stop full media assaults and misinformation campaigns against it etc.)
    1
  227.  @homeape.  but are you sure you can "vote people out of office"? how often has this happened, especially for the same party or even people not to return in a few years?
    1
  228. This is probably the actual reason why he is getting arrested - collaborating with "the enemy".
    1
  229.  @games4us132  if he accepts help offered by the Russian embassy, he may accidentally fall out of some window
    1
  230. Your sex life?
    1
  231. Typical black is white situation.
    1
  232.  @rifatno1  Right, the place ruled by a single guy, where spitting on the street or smoking a joint puts you in prison and get whipped. I also hear you can't be nude in your own apartment because someone could spy it out and snitch on you. Thanks, but no, thanks.
    1
  233. Are you kidding?
    1
  234.  @jozefkudrys1681  But Poland is quickly changing under the current EU shill government.
    1
  235. As a matter of principle, you don't install Microsoft products and don't purchase any hardware which collaborates with Microsoft (or Apple, Google, Facebook...).
    1
  236. Sam Bankman-Fried for stealing money outright from his greedy "customers".
    1
  237. In the West we still pretend. Russia/China do not even pretend any more.
    1
  238. Newbrak: all ultrarich people of this world are not really enemies, but colluding with each other. This also continues when they declare wars on each other (it's not like they have to fight in them themselves).
    1
  239.  @Petersilie161  When will you simpletons realize that one man's freedom of speech is another man's "terrorism"?
    1
  240.  @boredstudent9468  Right, you should arrest the train driver then if a criminal gets on a train with his stolen goods. After all he helped the criminal escape and did nothing to check his pockets. It is all reasonable to assume that many criminals travel on the train each day, so maybe also arrest the CEO of the train company. But let's not stop there. Who allows such evil train companies to operate? Obviously the politicians who make corporate laws in parliaments. So let's arrest those as well. Oops.
    1
  241.  azertyuiopqsdfghjklm  lol it shows
    1
  242.  azertyuiopqsdfghjklm  How can you be so sure the French police or secret services do what French law allows? And even if you are sure today, how can you be sure tomorrow and ever after?
    1
  243.  @Nina-cd2eh  I believe you are confusing accused with convicted (a subtle difference, I know)
    1
  244.  @DevinDTV  Governemnt and corporations work hand-in-hand against the common man.
    1
  245. The video outright mentione he has French citizenship.
    1
  246. Well, maybe you should take care of your children then, rather than letting Pavel Durov educate them.
    1
  247.  @Gargantuan_Inferno  Except that the "bullying" consits of words. Sticks and stones may break your bones, but words can never hurt you. By the way, teachers censoring unwelcome statements because of "bullying" is a nice example of an authoritarian system. But school is sort of designed to be authoritarian, in order to save effort for the teachers, not to benefit the students.
    1
  248. I would not say Telegram is your friend (if they wanted privacy, why the hell do they require phone number to sign up?), but having no privacy (as would be the case if any government controlled it) is even worse than having pretended privacy.
    1
  249.  @youknowwhatlol6628  Oryl? Then try divulging some of the government "secrets" online and see how long you can fly your flag.
    1
  250. wait until you hear the whole story of criminal and propaganda filled governments which want to ban telegram (and plant their propaganda and criminal content there outright)
    1
  251. Already lost all respect during Macron's escapades in the pandemic.
    1
  252. If "the criminals" can find each other, participate and coordinate through public Telegram channels, what prevents "the good policemen" from spying on those criminals and catching them through the same means? And if the criminals only do it privately, what prevents the criminals from doing it through any other software (including normal encrypted email or file uploads) rather than Telegram?
    1
  253.  @rolletroll2338  Well, the people who want to eliminate your privacy are the enemy.
    1
  254. He will come out a different guy, just like Julian Assange. It's kinda like a being reborn in a Russian gulag or Chinese prison camp... just under friendly supervision of US secret service. When he wakes up still-president Bidon will give great praise to American's loyal allies.
    1
  255. I suspect the whole world is NOT happy... why?
    1
  256. Ah yes, Trusted Computing where only Microsoft and co defines what is "trusted".
    1
  257. It doesn't matter if a few find a way around it. The point is that the majority will not and will have to suffer the consequences. Worse yet, laws will be made to criminalize workarounds. So you will be able to "work around" it while risking prison just like you are able to "work around" the anti-drug laws today.
    1
  258. I believe the only point is to prove that you took the picture, not what the picture was. Aka universal tracking of camera users.
    1
  259. No, it can't. The metadata will be incorporated into pixels of the image.
    1
  260. Don't worry, only government-approved people will be allowed to alter your metadata. For example, when they need Scientific Evidence to put you away for some time.
    1
  261. The only hope is that this project's real intent is to steal tons of government (e.g. your) money on consulting fees and useless legislation.
    1
  262. It will be called freedom subscription. We already have a "democracy tax" in Germany (forced to pay for lying public media whether you use them or not).
    1
  263. You will be able to upload your own freedom (for a fee) or use the default freedom supplied by the platform.
    1
  264. No, your comment reveals that you are new to the world.
    1
  265.  @LillyAnarkitty  Don't forget our brave military dropping cluster bombs on yer head in some regions.
    1
  266.  @tasseratee  too bad when the "private companies" ARE in fact part of your government... governing your life... making rules for you
    1
  267.  @morkallearns781  We neither need nor want to "uniformly enforce content authenticity". It's a sick idea born in the government thug brains.
    1
  268. The point is to verify whether it came from government or corporation-approved hardware. They don't care about what is "original" or not as long as you were forced to use their hardware to produce it and they can find out your identity.
    1
  269.  @crowe6961  Yeah the reality of EU pushing laws rejected by 90% of the population according to polls on their own website.
    1