• 18 Posts
  • 146 Comments
Joined 6 years ago
Cake day: August 24th, 2019
  • CriticalResist8@lemmygrad.mltoLate Stage Capitalism@lemmygrad.ml•Sam Altman prepares ChatGPT for its AI-rotica debut
    7·
    5 days ago

    It’s an indictment of openAI. They’re a fossil at this point, they need to monetize but their bubble is so big they can barely put a dent in it. How are you going to raise literally the 1 billion dollar a month they spend? China is rolling out models at a fraction of the cost that reach the same level, and notably their companies survive on things other than their LLMs - and they make their models open source too. Like they’re really not concerned about any of these questions over there, they’re just making their models.

    The problem is if openAI lets go, their bubble will fly away without them. It’s very cutthroat right now in western tech, with the big 3 companies all trying to outcompete each other but failing to find ways to. And openAI doesn’t do anything but AI, they’re using the startup philosophy (just build product with investment money and we’ll eventually figure out a way to monetize that sticks) but at their size it just doesn’t work anymore. Remember when they rolled out miniGPTs that you were supposed to be able to monetize?

    So any little advantage they can cling onto they will push, and this is how openAI has become a joke. I can’t remember the last time I used chatGPT, maybe once or twice a year if I want it to run python code directly but that’s about it.

    Oh yeah and z.ai came out of nowhere too last year and can build full stack apps for you. Well, they’re very buggy lol. But it gets a LOT of the job done and I can only expect it will become better down the line. Oh yeah and Chinese models are all free, even in the cloud version.

    A lot of this tech also relies on open-source contributions. For example the ubiquitous chatgpt UI style (message panel in the middle of the screen, chat history on the left) has been made into OpenWebUI and this is what most other providers use, possibly with their own tweaks to it but they start from that github repo.

  • CriticalResist8@lemmygrad.mltoLate Stage Capitalism@lemmygrad.ml•Do wellness drinks really do what they say?
    7·
    2 months ago

    Generally speaking when it comes to any supplement (drinks included) you can look at two things:

    1. The concentration of the ingredient
    2. Which way they get the ingredient in

    The first one is easier if you’re anywhere that’s not the US, since brands need to detail the ingredients list. Basically there’s doses wehre these ingredients start to work. You can look them up online generally. Many of these brands, especially in the energy drinks boom, underdose the active ingredients just so they can put the name front and center.

    The second thing can be trickier but you can also generally look it up. Notice it says lion’s mane extract for the Trip can, but only Ashwagandha. This is a particularly easy example to look at. The extract is good, it’s what you want. It’s essentially the concentrated active component without the not-interesting bulk - that is, every other molecule that forms fungi, such as the cells.

    But they don’t say that for the Ashwagandha, which implies they use the whole root, including the cellulose or whatever

    This logically lowers the concentration of the chemical that you want from Ashwagandha for this purpose. Whether the chemical actually works is another question, but this should be sufficient to say that it’s not going to be present in high enough concentrations if they don’t specify extract or the particular molecule.

    On the flip side you also have different binding agents. Magnesium is a big one. you can get magnesium oxide which is the cheapest form, but is also almost entirely rejected by the body. This means out of say 100mg of magnesium oxide, you will only process and absorb 10mg. Magnesium glycinate, bound to the glycine amino acid, is much more readily bioavailable to our body and the one you should take as a supplement because we digest glycine, but not oxide.

    Energy drink companies love the oxide form though because it’s cheap af since you can barely do anything with it, but they can still say there’s magnesium in their formulation, and sell you that can at a premium. They make huge margins on sales that’s why there’s an energy drink boom right now.

  • CriticalResist8@lemmygrad.mltoLate Stage Capitalism@lemmygrad.ml•To skirt copyright law, Anthropic bought and destroyed millions of physical books to train its "Claude" LLM
    151·
    3 months ago

    I don’t know if there’s anybody who hasn’t come to the same conclusion lol but ultimately, after (re-)reading and retyping my thoughts over and over, I come to two conclusions:

    This is a problem of capitalism but it’s also not saying much. The crux of the matter is copyright law and competition.

    On the one hand copyright law is so backwards and outdated (thank Disney) that the only way they could do this was to discard the books after scanning them. Cutting books, known as destructive scanning, used to be for a long time the only viable way to digitized books. With new methods however you can certainly do it without destroying the book, but many of these methods are patented.

    The other side of the coin is that these AI companies “need” to put out better, faster models all the time to stay in competition. It’s a fast-evolving industry with similarly cut-throat competition. If you fall behind, people stop using you and you don’t find funding.

    In higher-stage socialism, all of this would have basically been prevented. The SOE(s) responsible for AI research would have been told to preserve books even if it takes longer, or even find more efficient ways to train their models. There also wouldn’t be such a rush to put out marginally better models just to stay at the cutting edge. deepseek showed it’s possible to get a good model based on an original “cutting edge” model. Which means you only need the cutting edge model once, then you can decline it differently.

  • CriticalResist8@lemmygrad.mltoLate Stage Capitalism@lemmygrad.ml•what to hate about AI
    3·
    4 months ago

    the code is often over weirdly overengineered

    With the GPT + deepseek combo it fixes that problem (in my layman’s eyes); deepseek has a problem of overengineering especially if you don’t scope it correctly but if it works on an initial script, it will mostly stay within it. It’s also able to simplify GPT code.

    For the life of me I have never been able to learn javascript, so when I need to code something for ProleWiki I just throw the problem at them. You can look into https://en.prolewiki.org/wiki/MediaWiki:Common.js our script file for what they came up with - the color scheme picker, the typeface picker, and the testimonials code. Color picker is a bit overengineered because we have a theme that should only be available for one month out of the year and it was simpler to hardcode it in.

    I still have to think about the problem, if I don’t scope it well enough they will come up with whatever they feel like. For the color schemes when I had the first custom theme working I told them to refactor the code so that we could add new themes in an array, and they came up with everything. So now I can add unlimited themes by just adding a line in the array.

    Potential simplifications that they don’t think about (and hence why it’s good to know how the generated code works) is that there’s no need for className and LabelId in the array, it could generate that from the value. But eh, it works.

    edit - using it as a “copilot” is also a good use, though I find that sometimes it just utterly fails. It’s still an RNG machine at heart. But if you get it to work, it can really help unlock your other skills and not get stuck on one part of the process.

  • CriticalResist8@lemmygrad.mltoLate Stage Capitalism@lemmygrad.ml•what to hate about AI
    121·
    4 months ago

    I mean, AI as a word has been used in tons of different ways. We still say “NPC AI” in video games and that’s just a whole bunch of if statements, no LLM involved. And on the other end of the spectrum we still talk about AI in movies like I, Robot, with fully sentient machines. My line when I say “AI” with no qualifier is neural networks, the parameters that we hear so much about.

    And I don’t think GenAI is “the” grift like the tweet implies, because the AI they describe (machine learning, ML) is the exact same - neural networks with trained models. They talk about the Kinect using ML - you can do machine learning without the neural network - but was the Kinect not “wasteful”, “unethical”, and “useless”, to use their words? It was an expensive system that worked okay (on some tech demos) but barely had 3 games. The EyeToy for the PS2 was more fun.

    A lot of the conversation around Generative AI surrounds image generation and I feel that’s harmfully reductive. It centers a lot around artists and the purity of their art (as if they’re the only people impacted by AI) when there’s so much more to talk about; GenAI can do code - there was a whole discussion around it here the other day, and maybe it’s not super great code, but it can do code nonetheless and for people who don’t code and need something, it gets the job done.

    Yes, there is also a whole lot of stuff you can do with AI without LLMs. In fact, I’m not sure how LLMs specifically became so ubiquitous because you can do neural network AI stuff without ever needing an LLM. I remember back when AI became big (2022 or so), China announced they’d used an AI to map the wiring on a new ship model. What took an engineer one year to do was done by the AI in 24 hours.

    GenAI including LLMs have hard limitations that I think are, conversely, overlooked. LLM AI will not do everything, but it can get you part of the way there. The grift is moreso tech companies trying to pretend their toy is a panacea. When asked about AI making stuff up in an interview, OpenAI’s CTO “well you know, it’s very human in that regard [emphasis mine], because when we don’t know something, what do we do? We make stuff up”. They admit it themselves that they have a bullshit generator. But when it works, it works - you can use GPT 4o or o4 or whatever the new model is called as a tutor, for example, for photoshop, guitar, or whatever other hobby you have. It works great! You can ask it any question you have like a tutor, instead of being limited to what the page cares to tell you about! And yes I could ask someone, but: a- people are not necessarily available the moment I have a question and b- google is crap now and if you ask on most forums they will tell you to google it. So chatGPT it is. We just have to take into account that it might be making stuff up to the point that you need to double-check, and that OpenAI clearly has no plans to fix that (not that they even could).

    For coding my choice nowadays is start with chatGPT then pass it over to deepseek once I have the prototype, it works great.

  • CriticalResist8@lemmygrad.mltoMemes@lemmy.ml•Checkmate dronies
    2·
    4 months ago

    I just recently learned that Methyl anthranilate, the artificial grape flavoring, also repels birds and is used heavily by the agro industry nowadays. I imagine because of all the other products they also use they don’t need the birds around anymore.

    You can also use old cd-roms tied to a long-ish string btw, the sparkling makes birds stay away.

  • CriticalResist8@lemmygrad.mltoMemes@lemmy.ml•Checkmate dronies
    9·
    4 months ago

    he locked up a lot of people during the cultural revolution sometimes for quite arbitrary reasons

    He was not the one who signed arrest orders or ran trials though, a lot of different people were involved in that. You can’t blame Mao as an individual for stuff that other people did. You can’t run an entire state with just one person.

    the backyard steel and culling of the sparrows

    Nobody in the 60s knew about the importance of sparrows in agriculture. Not even in the West. I think it’s easy to say “oh well duh of course don’t kill the sparrows” but who here among us is an actual farmer? Who here knew that sparrows ate more bugs than grain before it was told to them? I can barely grow a plant, I have no room to judge others when it comes to growing food.

    I just didn’t find it to be very relevant to me

    I’m an adult in Europe and I find Mao’s writing to be both relevant and applicable. But there is Mao the general and Mao the chairman. By the end of his life he was definitely saying some stuff that I don’t think even he believed in. But theory is an all encompassing body, and that is true in all fields. One couldn’t read one physics paper about gravity and then say “now I know how to launch a rocket to the moon”. I opened up my copy of the red book randomly and here’s one:

    “Take the ideas of the masses and concentrate them, then go to the masses, persevere in the ideas and carry them through, so as to form correct ideas and leadership – such is the basic method of leadership”

    It makes perfect sense to me, but that’s also because I have the associated baggage to understand what he means there and how that fits in not only to more of Mao’s writings but also in regards to other figures, the ‘best practices’ if you will of organizing.

    I also found it really interesting that deepseek would refuse to answer any questions on Mao

    The deepseek devs want it to be mainly used for math, coding, and other STEM applications for lack of a better word. There’s nothing wrong with that, in fact personally I think people should stop using LLMs as oracles so much and focus them on tasks instead. Deepseek produces great results (and all for free with no rate limits) if you give it some code to start with, because it needs proper framing of the project to avoid trying to overdo it. I usually start with chatGPT, have it do the first working version of the code, and then switch it to deepseek to finish it, and it works almost perfect on the first try.