For many years, the problem of deepfakes has mostly focused on disturbing topics such as job replacement by the creative class, the visual spread of fake news, or so-called face-swapping, where the faces of famous celebrities or ordinary people are grafted onto filmed scenes. from pornographic content. After a long period of moral panic over AI-generated images, Black mirror this pushes the discourse into much more vivid imaginary territories. The first episode of the new season, in fact, works on the interdependence between the physical and digital world, between everyday reality and fiction. In a brief introduction, we are invited to look into the life of Joan, forced against her will to fire one of her former friends in cold blood and cheat on her partner, even if only hypothetically, with her ex. As evening falls, on the couch in front of the screen with her partner, the protagonist rediscovers key passages of the day just spent in a television series created by the confusing platform Streamberry (a pseudonym for Netflix itself). Series Joan is terrible tells about the life of this unfortunate victim, who “checked in” without reading the informed consent agreement for the transfer of his data (as in the memorable episode South A park), has agreed to a very special approach: all his experiences can be recycled with the help of quantum computers and AI to give life to a series of super deepfakes in which everyone’s life is re-experienced by several characters located at different ontological levels, in a kind of continuous loop between the interpreter and interpretable. The loop is clearly closed when the object (Joan) meets the subject of the representation (Salma Hayek). The series shows us the possible consequences of unfair integration between platform content, AI, deepfakes, the celebrity world and the supposed central role of the user, justified by the need to extract and process data produced by the “real” reality. As illustrated in this story, deepfakes populate this new “hybrid” communication ecosystem (Barile 2022) characterized by:
- A figurative dimension that mixes the real and the virtual in various ways;
- Client-centric mode, personalizing the experience for each individual user;
- An “empathic media” feature that allows you to interpret the emotional aspect of the data;
- Synthetic creativity capable of re-inventing possible worlds starting from a series of historical data.
Ever since the newspaper has become a central category in media representation, from classic storytelling to modern video game storytelling in the metaverse, artificial intelligence has shown itself to be particularly voracious in its ability to absorb the experiential content of everyday people. , which are converted into data suitable for training AI applications. So real pornography is increasingly becoming pornography of everyday life.
A recent book by Graham Meikle, Deepfakes (trans. German Perlego, 2022), works from a point of view that is halfway between the archeology of the media and the analysis of the imaginary. In fact, these “radical technologies” (Greenfield 2017) are comparable to older technologies such as electricity, which only changed society after their mass adoption. Therefore, it makes sense to study them now, at the very beginning of their establishment on a global scale. The author proceeds from the culture of remixes as a principle that can explain deepfakes by comparing them with the history of other creative practices: from cultural jam prank to memetics. But deepfakes go far beyond what was once called a convergent culture.
The unlimited set of images, videos and texts that we have shared through social media… is the raw material by which AI systems are trained to recognize, classify and recreate images. With proper compaction, these networks are capable of generating completely new images: copies without the original (ibid., p. 6).
Fans of Jean Baudrillard (surprisingly not mentioned by Meikle) will resonate with the definitions of the simulacrum proposed in Simulacra and simulations (1981). If at that time few people understood the meaning of self-referential images generated by a numerical matrix without any external reference, today this idea becomes easily recognizable in the evolution of deepfakes. Meikle cites the launch of the first iPhone and its facial recognition system as an example of the large-scale proliferation of synthetic media (Meikle, 2022, p. 22). By this term, we can refer to “images, audio or video that are generated by artificial intelligence processes” (ivi, p. 24). The author is engaged in a long study of synthetic porn, precisely created by AI systems, which can take various forms, such as porn without the consent of the characters in the series. Avengers (ivi, p. 44) placed in a pornographic context that is incredibly successful in terms of searches and views on Pornhub. Or, in everyday terms, practice revenge porn he makes the same substitution, but this time with relatives, acquaintances, and/or frequented people. In both cases, the misogynistic vocation of some applications, such as FakeApp and DeepNude, is used to create and distribute sexual content without consent (Gosse & Burkell 2020).
While this may seem like an “extreme” case, synthetic porn actually uses the same logic as the official business of exploiting the image of a celebrity through multiple channels. Namely, the appropriation of his image, his editing in non-traditional contexts, and his promotion in the public space (Meikle 2022, p. 45). There is a significant number of platforms and applications operating in this sector, from Mrdeepfake, which is the most visited community with the most porn versions and nudity associated with famous stars such as Emma Watson (about 478 content), to CelebJihad, which collects porn content. or genuine nudity, associating them with “aggressive and illegal” comments (ivi, p. 68).
Sexual performance is classically characterized by at least two predominant aspects: the direct activation of desire and the technical effectiveness of its staging (Breton 1995). A scientist’s vocation that uses technology and video as support to show what is really wrong, thus removing the bewitching charm of illusion.
Pornography … adds dimension to the space of sex, makes it more real than the real thing – this is its lack of seductiveness (…). You see it too close, you can see something that you have never seen before (…) thanks to anatomical zooming, the dimension of reality is abolished, the distance of the gaze gives way to an instantaneous and exaggerated representation: sex in its pure state, the end of perspective space, which is also the space of the imaginary and fantasy, end of scene, end of illusion (Baudrillard, 1979, pp. 37-38).
If the old pornography was above all transparency, technical reproducibility and analytical vocation, then the game of masks introduced by deepfakes reintroduces an aura of uncertainty, background noise and mystery into the world of high-tech creativity dominated by artificial intelligence.
We could include a few shades of deepfakes that are also related to the world of memes. Politics is one of the areas in which deepfakes about the most significant events of the moment are spreading. Pictures of Trump taken by the American police, or the Pope turned into a trapper, are significant in how AI manages to read some elements of human emotion in order to transform them into a certain image layout. In a much more fake than profound way, AI lends a touch of glamorous sensibility to the protagonists of American politics, turning them into performers halfway between Elton John and the drag queens in the Make America Glamorous Again memorable film. Deepfakes emphasize and reinforce the already acquired “horizontal” of the communicative-spectacular system and contribute to the democratization of the images of both celebrities and ordinary people. Both of the others are likely to be modeled or reimagined in a story that is different from itself. The difference between memes created with Photoshop, like a false juxtaposition of two images, and a synthesis made with AI (Vaccari & Chadwick 2020), is becoming more and more noticeable. As in the case of the great abdication of Yevgeny Viktorovich Prigozhin, metaphorically immortalized with the engraftment of his face to face Forrest Gumpwhen, after kilometers of a meaningless marathon, he invites everyone home. Far from a hoax cyberattack on Putin announcing that Ukrainian troops are invading the Kursk, Belgorod and Bryansk regions, declaring martial law and urging the inhabitants to flee, in the style of Orson Welles’s War of the Worlds. Prigozhin himself, a few months before the invasion, spoke out against deepfake audio recordings, accusing neural networks of miraculously mixing true sentences with false ones that he would never have uttered (Bershidsky 2023). If, until the nineties, the same disorienting effect was achieved through morphing, which allowed multiple identities to be smoothly superimposed, as is the case with a music video. Black or white Michael Jackson, the current disorientation caused by deepfakes is still already radical, indicative of a kind of alien and/or primitive sensibility (think AI-generated videos, songs, ads). The simultaneous feeling of familiarity and strangeness evoked by these images is reminiscent of the kind of disorientation robotics scientists call the “uncanny valley,” which also applies to virtual influencers (Berryman et al. 2021, Barile 2022) as well as deepfakes.
There are lighter and more institutionalized options, such as the rejuvenation effects of Harrison Ford, who, in a kind of concerted deepfake, returns from the age of eighty to the splendor of youth through AI-mediated rejuvenation. This image makes us rethink the well-known aphorism of Baudrillard, according to which “the sadartificial intelligence, lies in the fact that he lacks cunning and, therefore, intelligence. If the idea of perpetuating youth or forcibly restoring it with the help of technical devices such asHomo Deus Harari (2018), is pornographic in its own way, the idea of narratively hiding this interference with a crystal skull spell is a way to restore the erotic dimension of artificiality. Thus, the magical seductive power of illusion, destroyed by technology, returns to fashion in its own way, precisely thanks to the activity of artificial intelligence.
Recommendations
Barrel, N. (2022), Communication in New Hybrid Ontologies: From Platform to MetaverseMilan: Bocconi University Press.
Baudrillard, J. (1981), Simulacra and simulations. NY:
Semitext(e).
Baudrillard, J. (1979). About seduction, Paris: Galilee; translation an. (1985). seduction. Milan: EU.
Bershidsky, L. (2023), “Bloomberg: Deepfake Revolution Could Lead to Media Resurgence”, The Washington Post, March 29.
Berryman R., Abidin K., Leaver T., Topography of Virtual InfluencersPaper presented at AoIR 2021: 22nd Annual Conference of the Internet Explorers Association.
Breton, P. (1995), Utopia of communication. The myth of the planetary villageTurin, UTET.
Gosse K. & Burkell, J. (2020), Politics and porn: how the news media characterizes deepfake problemsCritical research in the field of media communications.
Harari Yu.N., (2017), Homo Deus. A Brief History of the FutureMilan, Bompiani.
Meikle, G. (2022), Deepfakes. Cambridge, British politics. Italian translation deepfakes, Perlego, 2022
Vakkari, K., Chadwick, A. (2020), “Deepfakes and disinformation: Exploring the impact of synthetic political video on deceit, uncertainty, and news credibility,” Social Media and Society, March 1-13.