• theshatterstone54@feddit.uk
    link
    fedilink
    arrow-up
    2
    ·
    19 days ago

    Not necessarily. A lot of the harms disappear when everything goes open, which is what this person stands for, and what OpenAI was supposed to stand for.

    Open LLM + Open Training Data = Open AI

    Copyright and IP concerns disappear with an open dataset.

    Open models are inherently more trustworthy because of an obvious reduction in vendor lock-in.

    • skillissuer@discuss.tchncs.de
      link
      fedilink
      arrow-up
      1
      ·
      19 days ago

      Copyright and IP concerns disappear with an open dataset.

      i don’t think i’d agree with that, doesn’t matter if dataset goes open if content went there without consideration for authors

      also even things like thispersondoesnotexist were used to mass-create fake identities and such

      • theshatterstone54@feddit.uk
        link
        fedilink
        arrow-up
        1
        ·
        19 days ago

        Yeah, but something like that would be super easy to find and fix without going through lawsuits. And I’d argue the dataset creators would be far less likely to add copyrighted material to the training data when it’s all out in the open and they can be immediately made to remove and retrain the AI without that data.