Why do you have non-profits like the Open Source Initiative saying that LLMs should be allowed to violate copyright?

Why do you have non-profits like the Open Source Initiative saying that LLMs should be allowed to violate copyright? This is why the Free Software Foundation is better, by the way.
https://opensource.org/blog/why-datasets-built-on-public-domain-might-not-be-enough-for-ai

Ape Out Shirt $21.68

Black Rifle Cuck Company, Conservative Humor Shirt $21.68

Ape Out Shirt $21.68

  1. 6 days ago
    Anonymous

    >Open source
    The people who are for "open source", but not free software are just corporate shills. They view the movement as a way to extract labor from programmers without having to pay for it. Free software values user freedom and has a principled stance on the rights of the programmer and the licensing. They want to violate copyright because they want to profit off other peoples work. They know that if they have to obey copyright their AI will stagnate and the bubble will burst.

    • 5 days ago
      Anonymous

      Shut up commie

      • 5 days ago
        Anonymous

        >Commie
        >Being mad that the compensation I choose is not money, but rather work
        You are just too stupid to understand GPL. You are the communist for wanting others work for free.

  2. 6 days ago
    Anonymous

    It's enough for LLMs, the quality of data is more important than quantity that's why the current LLMs are moronic.

    • 5 days ago
      Anonymous

      No you don't understand more cores will save the world you gotta believe me bro we just need to give NVIDIA another billion and tune the parameters a little bit trust me on this I'm good for it bro just a little more on the cloud bill for two more quarters and we'll be in the green for sure I just know it and push comes to shove we can just have a reduction in force to balance the books and make up the difference we already have everything we need just gotta train train train

  3. 5 days ago
    Anonymous

    make everything older than a decade public domain and you've solved almost every issue regarding copyright.

  4. 5 days ago
    Anonymous

    Learning isn't theft

    • 5 days ago
      Anonymous

      >AI
      >learning

    • 5 days ago
      Anonymous

      distributing model which embeds copyrighted material that it can reproduce is same as distributing copyrighted material, you are just obfuscating it.

      • 5 days ago
        Anonymous

        Did you know a model is trained on more images than it has bits?
        On average, it saves one third of a bit per image.
        A pixel is 27 bits. That means an AI model can only save 1/100th of a single pixel of each image.
        If you think it can reproduce an entire artwork while only having the information of 1/100 of a pixel, you must be braindead.

      • 5 days ago
        Anonymous

        Did you know a model is trained on more images than it has bits?
        On average, it saves one third of a bit per image.
        A pixel is 27 bits. That means an AI model can only save 1/100th of a single pixel of each image.
        If you think it can reproduce an entire artwork while only having the information of 1/100 of a pixel, you must be braindead.

        Or think about it this way:
        You have a book with 100 empty pages.
        Your task: read 5,000,000,000 books and learn about writing literature. Everything that you learn must fit into the 100 empty pages.

        Are you telling me that you can embed 5,000,000,000 books in just 100 pages and reproduce them?

        Are you telling me that the content of 5,000,000,000 images (let's say 5,000 GB of data) can be encoded in a mere 4GB model?

        If you did that then congratulations, you just broke the laws of physics https://en.wikipedia.org/wiki/Sloot_Digital_Coding_System

        • 5 days ago
          Anonymous

          The laws of physics don't really exist, they are just a fiction we invented. As for the Sloot Digital Coding System, those who think it impossible just don't understand it. It was not a compression algorithm, instead it was an exploit that allowed one to store and retrieve arbitrary amounts of information, written into the fabric of reality. The bytes that were output by Sloot functioned as an address or key.

          • 5 days ago
            Anonymous

            Not him. And yet, no one was able to replicate anything similar. Why?

  5. 5 days ago
    Anonymous

    They're right. Copyright is pure landlordism. The notion of owning an idea and intellectual """property""" is laughable.

    • 5 days ago
      Anonymous

      It's a joke
      All of it
      Like Musk just patented the letter X?
      Seriously he now owns a fricking letter of the Alphabet?
      Lmao

  6. 5 days ago
    Anonymous

    openai already uses shit out of the public domain and have been since GPT3, they dont even hide it
    theyre out of data which is why microsoft is scrambling to try and record you

  7. 5 days ago
    Anonymous

    >Why do you have non-profits like the Open Source Initiative saying that LLMs should be allowed to violate copyright?
    Because China won't ask. They'll just do it and not give a frick. And then they win the market

Your email address will not be published. Required fields are marked *