The US Copyright Office offers creative workers a powerful labor protective.

  • @illusoryMechanist
    link
    English
    31 year ago

    Does this mean we can, indeed, train AI on ChatGPT/other AI outputs?

    • Natanael
      link
      fedilink
      English
      91 year ago

      FYI there’s studies showing that repeatedly training ML on ML outputs degrades quality sharply after 4-5 cycles

    • @Prager_U
      link
      English
      01 year ago

      That is a used method for training other LLMs, I believe OpenAssistant did this.

        • Natanael
          link
          fedilink
          English
          21 year ago

          It does after 4-5 cycles yeah, built in algorithmic biases gets amplified to a degree that it breaks the model