13 votes

Introducing the pervert’s dilemma: A contribution to the critique of deepfake pornography

18 comments

  1. [11]
    MimicSquid
    (edited )
    Link
    I... hm. Isn't this really just the argument for the immorality of all porn and sex work? Heck, even more broadly, all of society can be tarred with this brush, as it's all part of a system that...

    I... hm.

    ... the Deepfake phenomenon can be considered morally impermissible on the basis of its role in gender inequality. The consumption of Deepfakes is undeniably a highly gendered phenomenon, and arguably plays a role in the social degradation of women in society. Sexual fantasies are not.

    Isn't this really just the argument for the immorality of all porn and sex work? Heck, even more broadly, all of society can be tarred with this brush, as it's all part of a system that supports and continues the degradation of all sorts of minorities. I'm not disagreeing with the conclusion, really. It's just... am I misreading it? It feels like a conclusion of "deepfakes are bad because society is bad" is really facile, but that's what I'm getting.

    15 votes
    1. [6]
      NaraVara
      Link Parent
      I also don’t really see how this argument doesn’t implicate sexual fantasies. Moreover, I would think men would be equally, if not more put off by deep fake porn being made of them so it’s pretty...

      I also don’t really see how this argument doesn’t implicate sexual fantasies. Moreover, I would think men would be equally, if not more put off by deep fake porn being made of them so it’s pretty clear that it’s not the macro view that prompts people’s intuitive revulsion to it.

      5 votes
      1. [5]
        TemulentTeatotaler
        Link Parent
        If I understand the paper, the author argues sexual fantasies are different because they are neutral while porn/deepfakes are predominantly produced-for/consumed-by men: An...

        If I understand the paper, the author argues sexual fantasies are different because they are neutral while porn/deepfakes are predominantly produced-for/consumed-by men:

        In contrast to Deepfakes, sexual fantasies are not normally considered a gendered phenomenon, and there is no immediately identifiable MAS responsible for their existence.

        An unshared/undistributable deepfake was being considered the same as a sexual fantasy on the individual level, but wrong on the societal level for reasons @eladnara elaborated on:

        the individual action of creating a Deepfake video (under conditions i and ii) is as morally permissible as a mere fantasy, while the phenomenon—the MAS as considered on a lower LoA—is to be deemed impermissible

        3 votes
        1. [4]
          NaraVara
          Link Parent
          Predominantly doesn’t mean “exclusively,” though. And even if it’s consumed by men that doesn’t mean it’s only women who are the subjects. What’s more, all of that is true of regular old porn too....

          Predominantly doesn’t mean “exclusively,” though. And even if it’s consumed by men that doesn’t mean it’s only women who are the subjects. What’s more, all of that is true of regular old porn too.

          Frankly I don’t think this author really understands how moral philosophy generally works. It’s not coming up with a legalistic definition to cover something you think is bad, it’s investigating why the thing feels like it’s bad and what that says about the broader moral truths insofar as we can posit they exist. He doesn’t seem to be looking into any aspect of the subjective experience of deepfakes feeling violating. Nobody is feeling violated by them because of some statistical notion around the gender balance of who is producing or consuming it. And men can also feel violated by it in ways the author doesn’t seem to understand or respect in the slightest.

          6 votes
          1. [3]
            Gaywallet
            Link Parent
            That is an incredibly uncharitable read. The article is focused on disparities for a good reason but it doesn't even introduce gender until a decent portion of the way through and in the context...

            He doesn’t seem to be looking into any aspect of the subjective experience of deepfakes feeling violating. Nobody is feeling violated by them because of some statistical notion around the gender balance of who is producing or consuming it. And men can also feel violated by it in ways the author doesn’t seem to understand or respect in the slightest.

            That is an incredibly uncharitable read. The article is focused on disparities for a good reason but it doesn't even introduce gender until a decent portion of the way through and in the context of the problem as it currently exists.

            But perhaps I'm misunderstanding what you're claiming here - is there a specific way in which men can feel violated by it that is not presented or is misunderstood in the paper? Keep in mind that even in the section which is heavily gendered, the author uses abstractions such as "A" and "B" referring to people for which gender is not specified.

            3 votes
            1. [2]
              Fal
              Link Parent
              Having only skimmed the article, there's a chance that there might be some kind of disconnect in the point the article gets across as opposed to the conclusion, but seems to suggest that the...

              Having only skimmed the article, there's a chance that there might be some kind of disconnect in the point the article gets across as opposed to the conclusion, but

              "...as isolated cases unrelated to other processes in society—there is no reason why Deepfakes should be deemed more morally impermissible than sexual fantasies. However, when the dilemma is considered on a low LoA—i.e., when we consider the truly morally relevant information—the Deepfake phenomenon can be considered morally impermissible on the basis of its role in gender inequality."

              seems to suggest that the author doesn't view deepfakes as unethical or immoral in and of themselves, but because of, well, its role in gender inequality. Maybe I just misinterpreted your comment, though.

              3 votes
              1. Gaywallet
                Link Parent
                I believe the author is arguing at the highest LoA at which it is impermissible. Much in the same way that the idea that murder can be argued against at a very high level LoA, the thrust of the...

                I believe the author is arguing at the highest LoA at which it is impermissible. Much in the same way that the idea that murder can be argued against at a very high level LoA, the thrust of the article is about the ways in which we can identify ethical problems at the highest level possible. In the case of sexism, gender norms, and other societal level LoAs, there's a clear negative consequence and argument against deep faked porn.

                However, the author spends time explaining that LoAs can exist on any level and makes sure to give examples of how this can run infinitesimally small. The idea that deepfakes cannot be found to be unethical or immoral at a different level for different people seems to ignore the LoA framework as outlined. To spend the article talking about every possible minute level of LoA in which it would be unethical seems like a fruitless endeavor.

                2 votes
    2. [4]
      TemulentTeatotaler
      (edited )
      Link Parent
      I think it's an argument that the emergence of Porn+ in a society that still has some unresolved problems that will be magnified by that is what people find moral fault with, while finding private...

      Isn't this really just the argument for the immorality of all porn and sex work?

      It feels like a conclusion of "deepfakes are bad because society is bad" is really facile

      I think it's an argument that the emergence of Porn+ in a society that still has some unresolved problems that will be magnified by that is what people find moral fault with, while finding private fantasies permissible.

      Maybe a different scenario would involve the invention of a weapon that increases an individuals lethality by 1,000x. The moral objections would be because it will lead to lots more deaths, not because of anything focused on individual use..?

      3 votes
      1. [3]
        MimicSquid
        Link Parent
        Ok, but then by that line of thought aren't all new things bad, in that they're the fruits of the existing paradigm?

        Ok, but then by that line of thought aren't all new things bad, in that they're the fruits of the existing paradigm?

        3 votes
        1. [2]
          TemulentTeatotaler
          Link Parent
          So, upfront, I thought the paper was interesting but I'm not really on-board with it. I'm probably not the best person to represent it. It's very narrow in scope (morality of a perfectly isolated...

          So, upfront, I thought the paper was interesting but I'm not really on-board with it. I'm probably not the best person to represent it.

          It's very narrow in scope (morality of a perfectly isolated deepfake) in a way I don't think maps on to reality and it doesn't make much of what I think is the core bit of deepfakes (their verisimilitude). I was expecting more of a real vs. symbolic focus when it first introduced layers of abstraction.

          There are already thirsty/threatening comments on photos of women in burkas, "bubbling porn", or photoshops. I think these introduce the same issues of making women feel uncomfortable with what their likeness may be used for, in concept, if not in degree.

          Getting beyond that, though, I don't think it follows that all new things would be bad. It's just a call to evaluate societal impact of something new, given the societal context its emerging in.

          Deepfakes can be used to turn photographs of dead relatives into video, for creative works, or perhaps for paleontological/historical/etc. research. It just happens to mostly be used for making porn out of people who didn't consent to that, with a dash of spreading political propaganda.

          The cotton gin could've been great in a time and place where it didn't make slavery more profitable. Cheap bioprinters and innovations in protein folding could be on the whole amazing or result in uncontrollable domestic bioterrorism. Quantum computing could be amazing or lead to a temporary breakdown of secure communication (e.g., online banking). Or a complicated combination where it depends on the "layer of abstraction" you're looking at.

          The paper's main argument was you can have different levels at which something can be evaluated as morally good or bad (e.g., mother/worker/citizen) and actions by an individual that are fine on their own but bad as a group. That resolved their "pervert's dilemma".

          2 votes
          1. MimicSquid
            Link Parent
            Fair enough. I suppose I like the general theory of Levels of Abstraction as a method of determining the morality of a situation, don't mind its application to the question of deepfakes, and feel...

            Fair enough. I suppose I like the general theory of Levels of Abstraction as a method of determining the morality of a situation, don't mind its application to the question of deepfakes, and feel like this paper in particular was barely dipping its toe in the murky waters of the situation.

            3 votes
  2. [7]
    mrbig
    Link

    Although to most people, Deepfake Pornography is intuitively unethical, it seems difficult to justify this intuition without simultaneously condemning other actions that we do not ordinarily find morally objectionable, such as sexual fantasies. In the present article, I refer to this contradiction as the pervert’s dilemma.

    2 votes
    1. [6]
      anothersimulacrum
      Link Parent
      I haven't yet read the rest of the article, so maybe this is addressed, but I'd say that the very clear difference between unethical activity and the ethical activity is that you can share and...

      I haven't yet read the rest of the article, so maybe this is addressed, but I'd say that the very clear difference between unethical activity and the ethical activity is that you can share and decieve people with the product of the unethical one, and can only share the description for the ethical one.

      4 votes
      1. [2]
        ShroudedMouse
        Link Parent
        The author maintains that even were it impossible to share the deepfake product, our moral intuitions would still be conflicted compared to an organic fantasy. Only half-way through myself but it...

        The author maintains that even were it impossible to share the deepfake product, our moral intuitions would still be conflicted compared to an organic fantasy.

        Only half-way through myself but it reads fairly well. It's the most scholarly address of Deepfakes (from an admittedly narrow perspective) that I've seen floatin around.

        4 votes
        1. TemulentTeatotaler
          Link Parent
          Using moral intuitions always feels a little shaky (and often breaks down on non-WEIRD populations), especially when introducing conditions that might not feel plausible, such as: I don't think...

          Using moral intuitions always feels a little shaky (and often breaks down on non-WEIRD populations), especially when introducing conditions that might not feel plausible, such as:

          The technology used by B guarantees that (i) A can never find out about the pornographic content in which A’s face is starring; and (ii) it is impossible to distribute the content to anyone else

          I don't think such a technology is possible, and I think that factors in to whatever intuition I have when considering it.

          4 votes
      2. [3]
        vakieh
        Link Parent
        Why does the ability to make it unethical? Surely that would be unethical if you made the choice to exercise that ability? (note this obviously doesn't apply to the release of tool with which to...

        Why does the ability to make it unethical? Surely that would be unethical if you made the choice to exercise that ability?

        (note this obviously doesn't apply to the release of tool with which to do it, just an individual privately using already existing tools)

        2 votes
        1. [2]
          eladnarra
          Link Parent
          So computer files can be copied easily, stolen through hacks, and even sometimes recovered after being deleted. When a person takes sexual photos of themselves, hopefully they are aware of those...
          • Exemplary

          So computer files can be copied easily, stolen through hacks, and even sometimes recovered after being deleted.

          When a person takes sexual photos of themselves, hopefully they are aware of those risks and both take precautions and accept the remaining risk. If someone makes a deep fake of them, there's no consent; they are stripped of the ability to make decisions about how much risk they want to take when it comes to depictions of them in sexual contexts. The person who made the deepfake made those decisions for them, which they didn't have the right to do.

          13 votes
          1. teaearlgraycold
            (edited )
            Link Parent
            It seems like there’s still a realm in which deep fakes are ethically safe. It’s just really high effort. The act of creating them today is high effort as well, but that might not always be the...

            It seems like there’s still a realm in which deep fakes are ethically safe. It’s just really high effort. The act of creating them today is high effort as well, but that might not always be the case.

            Beyond what’s ethical there is a matter of what’s right and wrong for yourself. If you go through someone’s Instagram, pull out as many pictures of their face as you can get, buy computer hardware to train a neural network on the photos, spend hours applying that network to a pornographic video of a carefully selected body double, and then render a final mp4 you have a serious problem. Then you need to make sure never to share the file and delete it, writing over the sectors with random bits

            That goes beyond sexual fantasy purely on an effort level. Someone performing that task is sexually obsessed with someone they can’t have. The target of their desire doesn’t know and won’t have to ever know. To anyone else it’s as if it never happened. But it’s not the best thing the deepfaker could have done for themselves.

            2 votes