all 91 comments

sorted by: hot top controversial new old
[–] 178 points 11 months ago (11 children)

I would not be surprised in the slightest if the cause of Nvidia drivers being absolute dogshit lately is because this absolute fucking MORON is forcing the devs to use AI to code the DRIVERS. THE ONE THING YOU DO NOT WANT FUCKING AI MEDDLING WITH.

If Microsoft did it with Windows 11 (which resulted in various SSD failures for a bunch of people), I have no doubt in my mind that Nvidia does it too.

  • source
  • hideshow 11 child comments
  • [–] 31 points 11 months ago (7 children)
  • [–] 79 points 11 months ago (2 children)

    In a word - instability. Downgrading driver versions for Nvidia cards is not an uncommon troubleshooting step now, which is not ideal.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 21 points 11 months ago

    That's binary blob drivers for you, you just try different versions and hope it gets better someday.

    One of the big advantages to open source drivers is that you can do a bisect to track some new breakage back to a specific patch. Sure, most people don't know how to do that, but there's a lot of people who can. And then the problem gets fixed for everybody.

  • source
  • parent
  • [+] 18 points 11 months ago (3 children)
  • [–] 3 points 11 months ago (2 children)
  • [–] 16 points 11 months ago (1 child)

    THE ONE THING YOU DO NOT WANT FUCKING AI MEDDLING WITH

    I mean... I can think of a lot more than just that one thing

  • source
  • parent
  • hideshow 1 child comment
  • [–] 2 points 11 months ago

    ...yeah, that list is way too extensive to mention, but I also said that because if you are programming something that literally makes your hardware be understood by the operating system, it should not have coding that is not created by a human.

  • source
  • parent
  • [–] 87 points 11 months ago (2 children)

    It wasn’t that long when Jensen was just the funny guy who’d come out and announced a bunch of cards and say some wacky shit, everyone outside of the tech press would ignore him and that was that.

    Kinda miss those days

  • source
  • hideshow 2 child comments
  • [–] 86 points 11 months ago (15 children)

    If AI does not yet work for a particular task, he said, employees should “use it until it does”

    Uh, using the AI doesn't train the AI, bud.

  • source
  • hideshow 15 child comments
  • [–] 33 points 11 months ago (11 children)

    Genuinely, this is the driving misconception people have about AIs right now. That somehow everybody using them is making them smarter, when really it’s leading to model collapse

  • source
  • parent
  • hideshow 11 child comments
  • [–] 4 points 11 months ago (10 children)

    Can you help correct this for me? Don't you feed them valuable training data and exposure to real world problems in the process of using them?

  • source
  • parent
  • hideshow 10 child comments
  • [–] 20 points 11 months ago (9 children)

    No. AI models are pre-trained, they do not learn on the fly. They are hoping to discover General Artificial Intelligence, which is what you are describing. The problem is that they don't even understand exactly how training even works. While engineers understand the overall architecture, the specific "reasoning" or decision-making pathways within the model are too complex to fully interpret, leading to a gap between how it works and why it makes a particular decision.

  • source
  • parent
  • hideshow 9 child comments
  • [–] 13 points 11 months ago (8 children)

    My assumption wasn't that they learned on the fly, it was that they were trained on previous interactions. Eg the team developing them would use data collected from interaction with model v3 to train model v4. Seems like juicy relevant data they wouldn't even have to go steal and sort

  • source
  • parent
  • hideshow 8 child comments
  • [–] 11 points 11 months ago

    That's true to an extend, but the interactions are only useful for training if you can mark it as good / bad etc (which is why sometimes apps will ask you if they were useful). But the 'best' training data like professional programming etc is usually sold at a premium tier with a promise not to use your data for training (since corporations don't want their secrets getting out).

  • source
  • parent
  • [–] 1 point 11 months ago (6 children)

    You can’t train ai on ai output. It causes degradation on the newly trained model.

  • source
  • parent
  • hideshow 6 child comments
  • [–] 7 points 11 months ago (5 children)

    First: that's wrong, every big LLM uses some data cleaned/synthesized by previous LLMs. You can't solely train on such data without degradation, but that's not the claim.

    Second: AI providers very explicitly use user data for training, both prompts and response feedback. There's a reason businesses pay extra to NOT have their data used for training.

  • source
  • parent
  • hideshow 5 child comments
  • [+] 0 points 11 months ago (5 children)
  • [–] 4 points 11 months ago (4 children)

    I mean - yeah, it is? This is a well-researched part of the data pipelines for any big model. Some companies even got into trouble because their models identified as other models, whose outputs they were trained on.

    It seems you have a specific bone to pick that you attribute to such training, but it's just such a weird approach to deny pretty broadly understood results...

  • source
  • parent
  • hideshow 4 child comments
  • [+] 1 point 11 months ago (4 children)
  • [–] 3 points 11 months ago (3 children)

    No, it doesn't. Unless you can show me a paper detailing that literally any amount of synthetic data increases hallucinations, I'll assume you simply don't understand what you're talking about.

  • source
  • parent
  • hideshow 3 child comments
  • [+] -2 points 11 months ago (3 children)
  • [–] 3 points 11 months ago (2 children)

    Yeah, that's what I guessed. Try to look into the research first before making such grandiose claims.

  • source
  • parent
  • hideshow 2 child comments
  • [+] 1 point 11 months ago (2 children)
  • [–] 1 point 11 months ago (1 child)

    Ah yes, and you can't show us that research because it goes to another school? And all companies that train LLMs are simply too stupid to realize this fact? Their research showing the opposite (which has been replicated dozens of times over) was just a fluke?

  • source
  • parent
  • hideshow 1 child comment
  • [+] 0 points 11 months ago (1 child)
  • [–] 1 point 11 months ago*

    no because this is literally in development, this isn’t some 60 year old mature tech

    Of course, you don't have research supporting your position because it's still in development. So obviously we can just ignore all the papers released over the last decade+ which show the opposite of what you're claiming - convenient!

    there is already academic accessible research talking about LLM issues of which the major concern is hallucinations, to the point where the word bailout is starting to make the rounds in the us from these very companies

    the argument is whether or not you believe this is inherent or fixable and a big focus is on the training

    anyone listening to any ai company right now is a damn fool with the obvious circular vendor bullshit going on

    Yeah, as I expected - you literally don't understand what this conversation is even about. Since you have a bone to pick with the industry, you make up random claims that you think make the industry look bad. But what you don't understand is: you're just making a fool of yourself by making subjective claims around topics you simply don't understand. Critique the AI industry for the greedy, useless shit they're doing and creating, not by making up wrong "facts" and ignoring all evidence against them.

    And just to save us both time, I'll list try to list positions you seem to think I hold, which I don't:

    • I don't think LLMs will ever get rid of hallucinations
    • I don't think LLMs will get better and better by only training on output from previous LLMs
    • I don't think LLMs are the path to AGI
    • I don't think any of the marketing done by AI companies is truthful

    If you choose to reply again and think I'm lying about not holding these positions, re-read the conversation until you understand it.

  • source
  • parent
  • load more comments (-1 replies)
  • load more comments (-1 replies)
  • load more comments (-1 replies)
  • load more comments (-1 replies)
  • load more comments (-1 replies)
  • [+] 60 points 11 months ago (1 child)
  • [–] 55 points 11 months ago (4 children)

    “I want every task that is possible to be automated with artificial intelligence to be automated with artificial intelligence,” he added. “I promise you, you will have work to do.”

    Well, yeah. Obviously, someone is going to have to unfuck everything AI fucked up...so, in a way, using AI is kind of like adding a layer of job security to your job.

  • source
  • hideshow 4 child comments
  • [–] 27 points 11 months ago (2 children)

    Reminds of when companies offshored their whole dev team and just sent requirements to them thinking they’d make code cheaper.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 27 points 11 months ago (1 child)

    Reminds of when companies offshored their whole dev team and just sent requirements to them thinking they’d make code cheaper.

    I mean, it was cheaper. It's just that it was also awful. It was basically like firing all your senior devs and giving their work to randos who can't code, but with plausible deniability.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 5 points 11 months ago

    Yes it was cheaper, even after they had to re-hire the folks they fired so they could fix the code they sent to a dev farm overseas with just a vague set of requirements and no oversight.

    But the same sort of attitude is happening with AI, like the code it’s making won’t require review and testing.

  • source
  • parent
  • [–] 3 points 11 months ago

    A blacksmith at the end of an assembly line. He looks at the assignment roster: 'Steak Knives'.

    The machine spews out broad swords and sometimes sporks, but they have two handles.

    Oh, there is a wooden stake that kind of looks like a knife.

    It's learning!

  • source
  • parent
  • [–] 111 points 11 months ago (3 children)

    “I want every task that is possible to be automated with artificial intelligence to be automated with artificial intelligence,” he added. “I promise you, you will have work to do.”

    I think every task that can be automated with LLMs is already automated. It's just that it's a very short list.

  • source
  • hideshow 3 child comments
  • [–] 67 points 11 months ago (2 children)

    Among them CEOs and some politicians.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 59 points 11 months ago (1 child)

    While I agree CEOs don't do much work. We still need a person there, because code can't be held accoun- nevermind, that ship has long sailed.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 42 points 11 months ago

    "A computer can never be held accountable. Therefore a computer must never make a management decision."

    • IBM motto, 1979

    Although at this point our computing philosophies from that era are ancient history according to these shortsighted people.

  • source
  • parent
  • [–] 41 points 11 months ago* (last edited 11 months ago)

    urged workers to keep relying on AI tools even when they fall short. If AI does not yet work for a particular task, he said, employees should “use it until it does”

    My honest reaction:

  • source
  • [–] 35 points 11 months ago (2 children)

    Honestly at this point I could kill these rich old fucks with my bare hands and feel nothing.

  • source
  • hideshow 2 child comments
  • [–] 28 points 11 months ago (1 child)

    The internal world these people live in must be insane.

  • source
  • hideshow 1 child comment
  • [–] 21 points 11 months ago

    Oh, it's all centered around the core concept that they're better than you, and consequences can't touch them.

    Now that you have a baseline of their mindset, some of their decisions.......are still really fucking stupid.

    Yeah. Nothing changed there. They just don't feel the consequences of their own destruction.

  • source
  • parent
  • [–] 24 points 11 months ago

    I will not believe these CEOs until they replace themselves with AI.

  • source
  • [–] 22 points 11 months ago (1 child)

    The output of this era will be remembered for its mediocrity but the shareholder value will be through the roof.

  • source
  • hideshow 1 child comment
  • [–] 19 points 11 months ago (2 children)

    “I promise you, you will have work to do.”

    Usually thats slang for, get ready to be laid off.

  • source
  • hideshow 2 child comments
  • [+] 38 points 11 months ago (6 children)
  • [–] 38 points 11 months ago (5 children)

    Jensen is actually pretty savvy, he knows exactly what it is and what's its purpose.
    It's whatever runs on Nvidia chips, and it's purpose is to sell Nvidia chips.

  • source
  • parent
  • hideshow 5 child comments
  • [–] 11 points 11 months ago (4 children)

    Yeah. It’s just the current band wagon he jumped on. Crypto was the previous one and gaming was the one before that.

    Anything to sell more chips.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 8 points 11 months ago (2 children)

    Seen like that it did help destroy the video game market. But with gamers cheering it on.

    Diablo 2, C&C, Red alert, Commandos, Point&Click, ... the best games are not about 3D graphics, fight me!

  • source
  • parent
  • hideshow 2 child comments
  • [–] 3 points 11 months ago

    Back when videocards were mainly for playing games, Jensen specifically was very explicit about his idea that games will not sell cards forever and they need to find a way to sell cards an mass for some other purpose (I still don't think he was right, but oh well). They did and still are doing a lot of very important scientific work, applying their stuff for research, from chemistry to astronomy, very cool stuff. Unfortunately, science pays less than virgins gambling on math equations, so that became their main focus after some time.
    But man, imagine the parallel world where we spend as much time and energy on science, as we do now on autocomplete that can fool idiots.

  • source
  • parent
  • [–] 36 points 11 months ago (2 children)

    This guy needs to fuck all the way off

  • source
  • hideshow 2 child comments
  • [–] 17 points 11 months ago

    Guess his golden parachute isn't quite ready yet.

  • source
  • [–] 48 points 11 months ago (4 children)

    Billionaire... For now.

    If Nvidia tanks he'll be a measly hunded-millionaire.

    P.s. large margin of error I have not deeply analysed Jensen Wang's stock holdings for a shitpost.
  • source
  • hideshow 4 child comments
  • [–] 11 points 11 months ago* (2 children)

    That's a pretty low net worth considering Nvidia's market cap.

    I assumed he could afford to start Jurassic Park for real by now, purely to skin a T-rex for his jacket collection.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 15 points 11 months ago (4 children)

    Reminds me of Ballmer taking iphones from his employees to try to make people use windows phone

  • source
  • hideshow 4 child comments
  • [–] 14 points 11 months ago (2 children)

    Someone should use AI to flood his company with emails, customer service requests, return requests, service tickets, etc. Just an endless avalanche of AI generated demands that must be dealt with immediately. And let it go on forever.

  • source
  • hideshow 2 child comments
  • [–] 1 point 11 months ago*

    better yet make every phone in every office, industrial park and warehouse ring. "Hi can I speak to Jensen? oh he's not there? will you take a message? FUCK OFF CLACKER!"

    ring ring over and over again, flood those voicemails and make them hang their head every time it rings.

  • source
  • parent
  • [–] 9 points 11 months ago

    Jensen is insane

  • source
  • [–] 8 points 11 months ago

    He is the insane one

  • source
  • [–] 16 points 11 months ago (2 children)

    Old people's opinions are often the result of failing biological hardware, or wetware if you prefer. He should focus on anti-aging instead of forcing people to use comical tools used to make farting fat men for anything serious or important.

    Maybe he should fly on an AI piloted airplane until it lands right.

  • source
  • hideshow 2 child comments
  • [–] 1 point 11 months ago (1 child)

    Planes actually have pretty sophisticated autopilot. I would like to see him fly on a plane piloted by an LLM though.

  • source
  • parent
  • hideshow 1 child comment
  • [–] 1 point 11 months ago* (last edited 11 months ago)

    Yes but I consider avionics to be almost quaint compared to the complexity of modern computers and software. A FADEC is still just a FSM, deterministic, and based on Boolean logic.

    From what I've seen of this new-fangled AI stuff all the kids are talking about, it seems the ideas of IF-THEN-ELSE or functions, or code itself, are about as relevant as rubbing sticks together to make smoke, not even a fire.

    Being able to run zillions of "parameters", whatever they are, on endless quillions of petabytes of "training data" to get ... something... is something that's beyond my ken.

    But I still think we should shove this old stinker in an AI-designed and flown plane. Have you read this novel:

    https://en.wikipedia.org/wiki/Incompetence_(novel)

    He didn't predict AI, but rather, a future that makes Idiocracy look preferable.

  • source
  • parent
  • [–] 5 points 11 months ago

    Nooo Not my Stock Portfolio!!! --Jensen Huang probably

  • source
  • [–] 5 points 11 months ago

    Old man yells at ai

  • source
  • [–] 2 points 11 months ago* (last edited 11 months ago)

    NVIDIA AI:
    busted

  • source