Robin Williams’ daughter Zelda says AI recreations of her dad are ‘personally disturbing’::Robin Williams’ daughter Zelda says AI recreations of her dad are ‘personally disturbing’: ‘The worst bits of everything this industry is’

  • scarabic@lemmy.world
    link
    fedilink
    English
    arrow-up
    35
    arrow-down
    11
    ·
    1 year ago

    This is such a random thought and I don’t mean to conclude anything by it:

    I’ll bet people felt this way about the very first audio recordings.

    How creepy to hear your sibling’s voice when that sibling is not even in the room!

    …and moving pictures:

    It looks like your mother is right there but she’s been dead for 10 years! Gah!

      • scarabic@lemmy.world
        link
        fedilink
        English
        arrow-up
        2
        ·
        1 year ago

        You tend to consent to a photo or video

        I’m not sure what you mean. There’s nothing more consensual about photography necessarily. Paparazzi are a thing, for example.

        I think the real difference here is that we understand video and audio recordings, we even have some laws governing when you can record someone. So we are comfortable with those technologies. Above all, we’re used to them.

        AI isn’t the exact same thing but I think the main source of discomfort is its newness and mysteriousness. We don’t have laws governing it. We don’t understand it very well. This makes it creepy.

        • Julius_Seizures@lemmy.world
          link
          fedilink
          English
          arrow-up
          6
          arrow-down
          1
          ·
          1 year ago

          I think consent is the most important discussion here. The people that continue to profit (monetarily or otherwise) off dead creators are often looked down upon, eg. Brian Herbert’s Dune continuation, Stephen Hillenberg’s death and continuation of spongebob (and it’s spin offs), etc. Terry Pratchett had in his will to use a steamroller to destroy all his unfinished works as he knew if not they would likely be used to profit after his death without him.

          I’m a proponent of the recent advances in machine learning, I use machine learning in my field and I write and use models for hobby level things. I’m also fully a proponent of using these things ethically, and consent here is the most important thing.

          If I created a doctored photograph of Robin Williams (even doing something innocuous) that was clearly not something he did and plastered it around the internet it would be in bad taste. If Robin Williams consented to people doing that then sure whatever its nbd. Photographs and recordings should be used with consent, and things like the paparazzi taking non consensual photos are not looked upon as particularly ethical endeavors.

          • scarabic@lemmy.world
            link
            fedilink
            English
            arrow-up
            2
            arrow-down
            2
            ·
            1 year ago

            Let’s say one of your parents dies and years later you stumble upon a voice recording that your sibling made of them. Your heart would probably be warmed just to hear their voice. It wouldn’t change that if you realized that your brother had recorded them from behind without their knowledge. You’d still be comfortable with that representation of your father.

            Another example: there are services which can take an old photo of a dead relative and turn it into a sort of “Harry Potter moving picture” kind of deal, using deepfake technology. Most people are amazed and touched in a positive way when they see these.

            I think someday when AI is much more mundane to us, someone out there will take old voice recordings of their long lost father, train an AI bot on them, and present it as a gift to their sibling. That sibling will have a conversation with it, and their eye will mist up, and they’ll say thank you this is so touching and wonderful.

            It’s merely a question of being comfortable with the technology itself.

    • Something_Complex@lemmy.world
      link
      fedilink
      English
      arrow-up
      22
      ·
      edit-2
      1 year ago

      To be honest it is a bit creepy if it wasn’t from Robin Williams’ personality.

      If you hear a message you brother left you is one thing. But listening to him taking when someone else is faking his voice and saying whatever they want.

      That’s the only difference, those video recording where of you brother.

      These deep-fake things are someone else speaking in your brother’s voice. A corporation using your brother to sell products and services.

      Nothing to do with him and his personality

      • Comment105@lemm.ee
        link
        fedilink
        English
        arrow-up
        14
        arrow-down
        1
        ·
        1 year ago

        Yeah, there’s a significant difference between a recording and generating.

        • MimicJar@lemmy.world
          link
          fedilink
          English
          arrow-up
          1
          arrow-down
          2
          ·
          1 year ago

          What are your thoughts on things like Photoshop? As an example https://old.reddit.com/r/Damnthatsinteresting/comments/psqkz6/how_you_remove_a_ex_from_your_family_picture_with/

          That is generating an image, and in turn an event that didn’t happen.

          There are also cases of “repairing” old photos. Sometimes an old black and white photo is torn or faded, but we can restore it, we can add back things that aren’t there.

          After someone passes away you often hear people say “I wish I could hear their voice again” or “I wish I could have one last conversation”.

          I’m not going to deny that there are A LOT of terrible things that could be done with this technology. I just wonder what positives exist and how we might improve things.

          • Comment105@lemm.ee
            link
            fedilink
            English
            arrow-up
            2
            ·
            1 year ago

            I guess the two categories didn’t cover the whole space of possibilities.

            But many cases are clear-cut.

        • scarabic@lemmy.world
          link
          fedilink
          English
          arrow-up
          2
          arrow-down
          5
          ·
          1 year ago

          I think the difference is that people understand recordings now and do not understand genAI yet. Therefore the former is “fine” and the latter is “creepy.” You could make many arguments about recordings that someone from the 1800s would be concerned about: Taking my words out of context. Editing my words to change what I said. Am I accountable for what I said when it is heard as a recording? Is my permission required for recording?

          As you’ll notice, we even have laws for some of this now. Those no doubt came from people flipping their shit about the new technology, just as we’re doing now.

    • eumesmo@lemmings.world
      link
      fedilink
      English
      arrow-up
      5
      arrow-down
      1
      ·
      1 year ago

      It’s not just a matter of discomfort for something new, but at something highly dangerous. Deepfakes have several bad and disturbing use cases, like itentity theft, sexual exploitation, marketing abuse, political manipulation, etc. In fact, I hard to find a significant good use of such technology.

      • stevedidWHAT@lemmy.world
        link
        fedilink
        English
        arrow-up
        1
        ·
        edit-2
        1 year ago

        Ops point remains, this is exactly what everyone said about photos and then videos and then video with sound etc.

        You’ve always been told you can’t see what’s on the internet, now that’s even more true.

        There are ways we process and handle new tech, there’s a grace period to figure out issues and solutions.

        Part of the problem is regressionist ideals holding everyone back from making real changes. Being able to generate nudies of your crush is the tip of the iceberg and demonstrates our ability to create teachable models that perform well and reliably to reconstruct images from noise. There lots of applications but ultimately making images is just art and it’s sorta hard to break out of that sphere easily.