4 ms·

It's not bending the knee, that's how they want it to be perceived, but what's really happening is that they're trying to pull up the ladder.

reaperman 3y ago
It'll be a temporary 10-year moat at best. Eventually consumer-grade hardware will be exaflop-scale.
- digging 3y ago
  A 10-year moat in AI right now is not a minor issue.
- classified 3y ago
  By then it will be legally locked down and copyrighted to hell and back.
  - reaperman 3y ago
    I'm sorry, what? - What is OpenAI's level of copyright now? - How is it going to be more "copyrighted" in the future? - How does this affect competitors differently in the future vs. the copyright that OpenAI has now?
    - JumpCrisscross 3y ago
      > What is OpenAI's level of copyright now Limited. They’re hoping to change that. It’s no secret that open-source models are the long-run competition to the likes of OpenAI.
      - reaperman 3y ago
        I don't understand what "Limited" entails. I was pointedly asking for something a bit more specific.
        JumpCrisscross 3y ago
        > don't understand what "Limited" entails Nobody does. It’s being litigated. They want it legislated. Model weights being proprietary by statute would close off the threat from “consumer-grade hardware” with “exaflop-scale.”
        reaperman 3y ago
        > "Nobody [knows what it means]" [re: knowing what 'limited' means] Then why did you say "Limited"? Surely YOU must have meant something by it when you said it. What did YOU mean? I don't think you're saying that you are repeating something someone else said, and you didn't think they knew what they meant by it, and you also don't know what you/they meant. Correct me if I'm wrong, but I'm assuming you had/have a meaning in mind. If you were just repeating something someone else said who didn't know what they meant by it, then please correct me and let me know -- because that's what "nobody knows what it means" implies, but I feel like you knew what you meant so I'm failing to connect something here. > It’s being litigated. I'm not able to find any ongoing suits involving OpenAI asserting copyright over anything. Can you point me to one? I only see some where OpenAI is trying to weaken any existing copyright protections, to their benefit. I must be missing something. I'm also unable to find any lobbyist / think-tank / press release talking points on establishing copyright protections for model weights. Where did you see this ongoing litigation?
        JumpCrisscross 3y ago
        These are broad questions whose answers are worth serious legal time. There is a bit in the open [1][2]. [1] https://www.bereskinparr.com/doc/chatgpt-ip-strategy https://www.bereskinparr.com/doc/chatgpt-ip-strategy [2] https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem https://hbr.org/2023/04/generative-ai-has-an-intellectual-pr...
        reaperman 3y ago
        Hmm, these links don't have anything about "model weights being proprietary". They also don't have anything about current litigation involving OpenAI trying to strengthen their ability to claim copyright over something. Where it does mention OpenAI's own assertions of copyright? OpenAI seems to be going out of their way to be as permissive as possible, retaining no claims: From [1] > OpenAI’s Terms of Use, for example, assign all of its rights, title, and interest in the output to the user who provides the input, provided the user complies with the Terms of Use. Re: [2]: I believe I referenced these specific concerns earlier where I said: "I only see some where OpenAI is trying to weaken any existing copyright protections, to their benefit. I must be missing something." This resource shows where OpenAI is trying to weaken copyright, not where they they are trying to strengthen it. It's somewhat of an antithesis to your earlier claims. I notice you don't have a [0]-index, was there a third resource you were considering and deleted or are you just an avid Julia programmer?

reaperman 3y ago

It'll be a temporary 10-year moat at best. Eventually consumer-grade hardware will be exaflop-scale.

digging 3y ago
A 10-year moat in AI right now is not a minor issue.
classified 3y ago
By then it will be legally locked down and copyrighted to hell and back.
- reaperman 3y ago
  I'm sorry, what? - What is OpenAI's level of copyright now? - How is it going to be more "copyrighted" in the future? - How does this affect competitors differently in the future vs. the copyright that OpenAI has now?
  - JumpCrisscross 3y ago
    > What is OpenAI's level of copyright now Limited. They’re hoping to change that. It’s no secret that open-source models are the long-run competition to the likes of OpenAI.
    - reaperman 3y ago
      I don't understand what "Limited" entails. I was pointedly asking for something a bit more specific.
      - JumpCrisscross 3y ago
        > don't understand what "Limited" entails Nobody does. It’s being litigated. They want it legislated. Model weights being proprietary by statute would close off the threat from “consumer-grade hardware” with “exaflop-scale.”
        reaperman 3y ago
        > "Nobody [knows what it means]" [re: knowing what 'limited' means] Then why did you say "Limited"? Surely YOU must have meant something by it when you said it. What did YOU mean? I don't think you're saying that you are repeating something someone else said, and you didn't think they knew what they meant by it, and you also don't know what you/they meant. Correct me if I'm wrong, but I'm assuming you had/have a meaning in mind. If you were just repeating something someone else said who didn't know what they meant by it, then please correct me and let me know -- because that's what "nobody knows what it means" implies, but I feel like you knew what you meant so I'm failing to connect something here. > It’s being litigated. I'm not able to find any ongoing suits involving OpenAI asserting copyright over anything. Can you point me to one? I only see some where OpenAI is trying to weaken any existing copyright protections, to their benefit. I must be missing something. I'm also unable to find any lobbyist / think-tank / press release talking points on establishing copyright protections for model weights. Where did you see this ongoing litigation?
        JumpCrisscross 3y ago
        These are broad questions whose answers are worth serious legal time. There is a bit in the open [1][2]. [1] https://www.bereskinparr.com/doc/chatgpt-ip-strategy https://www.bereskinparr.com/doc/chatgpt-ip-strategy [2] https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem https://hbr.org/2023/04/generative-ai-has-an-intellectual-pr...
        reaperman 3y ago
        Hmm, these links don't have anything about "model weights being proprietary". They also don't have anything about current litigation involving OpenAI trying to strengthen their ability to claim copyright over something. Where it does mention OpenAI's own assertions of copyright? OpenAI seems to be going out of their way to be as permissive as possible, retaining no claims: From [1] > OpenAI’s Terms of Use, for example, assign all of its rights, title, and interest in the output to the user who provides the input, provided the user complies with the Terms of Use. Re: [2]: I believe I referenced these specific concerns earlier where I said: "I only see some where OpenAI is trying to weaken any existing copyright protections, to their benefit. I must be missing something." This resource shows where OpenAI is trying to weaken copyright, not where they they are trying to strengthen it. It's somewhat of an antithesis to your earlier claims. I notice you don't have a [0]-index, was there a third resource you were considering and deleted or are you just an avid Julia programmer?