• Sanyanov@lemmy.world
    link
    fedilink
    English
    arrow-up
    15
    ·
    edit-2
    1 year ago

    Afaik we’re still yet to find out whether viewing AI-generated material makes an individual look for real-life child abuse imagery.

    I believe viewing the latter allows many to keep real-life urges under control (might re-check materials on that), but it obviously comes with its own issues. If we can make AI generated child pornography, and if it doesn’t make people go look for the “real stuff”, we might actually make a very positive impact on child safety.

    • Uriel238 [all pronouns]@lemmy.blahaj.zone
      link
      fedilink
      English
      arrow-up
      8
      ·
      1 year ago

      According to the few studies we have in the nineties and aughts most people who have sexual attractions to kids are aware acting on them can be harmful and will turn to alternative ways to explore them (when they can’t be suppressed or redirected.) So yes, now we have victimless ways to produce porn, the objections are to the fetishes themselves, not to resulting violent behavior.

      That said people commonly and openly express their distaste for such people, more so than domestic violence offenders who assault their kids, just not sexually. The same general disdain for child-sex-attracted adults does not transfer to action to feed children or protect them from abusive households.

      That said, when we’ve worried about fiction driving people to act it out in reality, historically this has been demonstrated wrong every single time. Women are not driven to scandal and harlot behavior from trashy romance. Teens are not driven to violence from violent movies or video games. We can expect porn featuring childreb is not going to compell someone to seek to actually sex-assault kids.

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

        It would take a lot of shots to make a meaningful change in the database.

        It would probably require training on existing data, which by itself is questionable, but I lean to the side that it might be worth it for the cause.