• 0 Posts
Joined 4 years ago
Cake day: June 7th, 2020

  • If a pedophile creates a hospital/clinic room setting and photographs a naked kid, will it be okay? Do you understand these problems impossible to solve just like that? Parents also take photos of their kids, and they do not take photos like a doctor would. They take photos in more casual settings than a clinic. Would parents be considered pedophiles? According to the way you propose to judge, yes.

    You are basically implying that social defamation is what matters here, and the trauma caused to victim of such fictional media is a problem. However, this is exactly what anti-AI people like me were trying to warn against. And since these models are open source and in public hands, the cat is out of the bag. Stable diffusion models work on potato computers and take atmost 2-5 minutes to generate per photo, and 4chan has entire guides for uncensored models. This problem will be 100x worse in a couple years, and 1000x worse in the next 5 years. And infinitely worse in a decade. Nothing can be done about it. This is what AI revolution is. Future generations of kids are fucked thanks to AI.

    The best thing one can do is protect their privacy, and photos from being out there. Nobody can win this battle, and even in the most dystopian hellhole with maximum surveillance, there will be gaps.

  • Have you considered the problem of doctors, married parents and other legitimate people being labelled as CSAM users and pedophiles? This has already happened, and they are not obligated to take the brunt of misjudgement of tools developed to judge such media. This is not a hypothetical scenario, and has already happened in real world, and has caused real world damage to people.

    The argument of planted CSAM is not incoherent, but has also played out with many people. It is one of the favourite tools for elites and ruling politicians to use. I am less worried about it because such a law thankfully does not exist, that will misjudge the masses brutally for fictional media.

  • An image is not merely an arrangement of pixels in a jpeg,

    I am not one of those “it’s just pixels on a screen” people. But if it was not recorded in real world with a camera, it cannot be real.

    Who will be the judge? If there is some automated AI created, who will be the one creating it? Will it be perfect? No. We will end up in the situation that Google caused to users, like doctors, married parents and legitimate people being labelled as pedophiles or CSAM users. It has already happened to me in this thread, and you also said it. The only accurate way to judge it will be a very large team of forensic experts on image/video media, which is not feasible for the amount of data social media generates.

    not every law needs to have a perfectly defined line

    And this is where the abuse by elites, politicians and establishment starts. Activists and dissidents can be easily jailed by CSAM being planted, which would in this case be as simple as AI pictures being temporary drive by downloads onto target’s devices.

  • It was not a threat, but a hypothetical example to gauge the reaction of that reactionary baiter.

    The problem with claiming AI generated art as CSAM is that there is no possible way to create an objective definition of what “level” of realism is real and what is not. A drawing or imaginary creation is best left not defined as real in any capacity whatsoever. If it is drawn or digitally created, it is not real, period. Those people thinking of good uses of AI were too optimistic and failed to account for the extremely bad use cases that will spiral out of control as far as human society goes.

    Even though China is incredibly advanced and proactive on trying to control this AI deepfake issue, I do not trust any entity in any capacity on such a problem impossible to solve on a country or international scale.

    I just had a dejavu moment typing this comment, and I have no idea why.