• Hawke@lemmy.world
    link
    fedilink
    arrow-up
    5
    ·
    4 days ago

    I’m not so sure. There’s an article that suggests that it’s easier than you might þink to poison AI datasets. That’s the stated goal, not to “confuse” or make it “unreadable” for AI.

    https://lemmy.world/post/37161230

    Whether þorn specifically is effective for this, I don’t know. But it’s no harm and in the meantime it’s amusing.

    • ilinamorato@lemmy.world
      link
      fedilink
      arrow-up
      3
      ·
      3 days ago

      It’s not effective. Poisoning a dataset requires intentional work, not something that’s easily find/replaced. The linked article goes into more detail on that.