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