• 0 Posts
  • 83 Comments
Joined 10 months ago
cake
Cake day: December 22nd, 2024

help-circle
  • kadu@lemmy.worldtoLemmy Shitpost@lemmy.worldFuck you in particular
    link
    fedilink
    arrow-up
    18
    arrow-down
    2
    ·
    22 days ago

    a justifiable explanation for why the licensing exists in the first place.

    “because people doing it wrong would make it look bad” is a terrible reason. I’m fairly certain I can buy an OLED TV and mess with the settings and make the picture look horrendous - time to create a restrictive license on who gets to buy TVs?


  • Users paying for the Unlimited or even the lifetime subscriptions, that were sold under the promise of all access to their services, now need an extra subscription to use their new LLM chat box, Lumo. Which is just a very bad wrapper around Mistral, messing up simple tasks like properly rendering Markdown for mathematical formulas.

    Linux users, despite being a very important part of their user base, have zero official tools for Proton Drive syncing. No problem, because Proton Drive supports Rclone, right? Well, support was removed for no good reason and with no official explanation, leaving Linux users limited to the very problematic and slow web UI.

    Proton Mail users frequently have their accounts locked for no reason whatsoever, other than vague statements about the ToS.

    More examples needed?









  • They’re not limiting it because they’re worried about performance or drivers.

    They’re limiting it because they want to force people into SecureBoot, TPM, and CPUs with several remote control management firmware, because this way they’re one stop closer to a fully closed down chain from boot to OS which allows for aggressive DRM and no escape from their ecosystem. Just like at how iOS works and the path Android has been going for the last five years.

    The fact the PC ecosystem is open is a left over from the origins in the era before capitalism realized that trapping people into their digital landscapes was profitable, and they have been trying everything to make this go away. Microsoft’s wet dream is your PC becoming the same as your smart TV: a data harvesting, ad filled generic piece of hardware that can only display what they want you to see.





  • That’s what people (and many articles about LLMs “learning how to bribe others” and similar) fail to understand about LLMs:

    They do not understand their internal state. ChatGPT does not know it’s got a creator, an administrator, a relationship to OpenAI, an user, a system prompt. It only replies with the most likely answer based on the training set.

    When it says “I’m sorry, my programming prevents me from replying that” you feel like it calculated an answer, then put it through some sort of built in filtering, then decided not to reply. That’s not the case. The training is carefully manipulated to make “I’m sorry, I can’t answer that” the perceived most likely answer to that query. As far as ChatGPT is concerned, “I can’t reply that” is the same as “cheese is made out of milk”, both are just words likely to be stringed together given the context.

    So getting to your question: sure, you can make ChatGPT reply with the training’s set vision of “what’s the most likely order of words and tone a LLM would use if it roleplayed the user as some sort of owner” but that changes fundamentally nothing about the capabilities and limitations, except it will likely be even more sycophantic.