By CAMILO SOARES*
The lack of affection in a dehumanized gaze leads us to a dangerously smooth and predictable, generalist and idiotic world.
“I think we are reaching the end of times. Humans are losing faith in themselves.”
(Hayao Miyazaki).
Between the rifle and the machete
The experience of visiting the photography exhibition Gold – Serra Pelada Gold Mine, by Sebastião Salgado, at Caixa Cultural in Recife was marked by a serious error in the audio description, probably generated by artificial intelligence. The description distorted the content of an emblematic image, exchanging a rifle for a machete and inverting the roles between oppressor and oppressed. This event reveals not only the risk of dehumanization in the uncritical use of artificial intelligence, but also a lack of respect for the audience and the original work.
When faced with the lost importance of affection and context for the reading of artistic images by artificial intelligence, we point out the dangers of a generalist perception promoted by technology, in addition to the authorial and environmental problems that emerge in cases such as the Ghibli trend, from ChatGPT.
Even preserving a healthy critical eye, whenever we come across an exhibition like Gold – Serra Pelada Gold Mine, by Sebastião Salgado, we are convinced that we have entered the work of an artist. We are aware of the time it took for a look to mature and, more evidently, of the time the photographer took to understand and capture a series of images capable of expressing what he felt about that reality, where, at the time, thousands of men huddled like ants in a 200-meter-deep pit in the Serra dos Carajás, Pará, dehumanizing themselves daily in search of luck and fortune in the form of gold.
Although nothing is very new in a series made in 1986 (despite the curatorship stating that 31 of the 56 photographs are previously unpublished, changing little or nothing in the appreciation of the set) the opportunity to see or re-see an exhibition of this size is always an invigorating experience for attentive eyes, lovers or simply curious.
However, what stole the show during the visit to Caixa Cultural in Recife was not the mastery of the framing, the evident social approach or even the unfortunate printing medium – which is far from being the best way to express the nuances of the silver grains of the photography that Sebastião Salgado was producing at the time; what really stole the show was an invisible detail, but one that reflects, in a certain way, the dangers of our times. When listening to the audio description comments, I realized that a gross error, one of those that make you feel ashamed, could only be the result of artificial intelligence.
I stopped in front of one of the emblematic photos of the series, which I searched for in the tangle of images as if searching for traces of an old memory. This richness, at once temporal, spatial and affective, gives the image a power as mysterious as it is revealing, with which Gilles Deleuze, crossing the Bergsonian cone of memory, seems to open up a map for us to understand, in this movement between intimacy and exteriority, the secrets of the image: “We do not move from the present to the past, from perception to memory, but from the past to the present, from memory to perception.” (DELEUZE, G. Bersonism, p.67, apud LISSOVSKY, 2008 p. 107).
The photo captures a strong and fearless prospector, dressed in rags, holding the rifle of a white soldier, who seems so small and destitute in front of that black and muscular body fixed to the mountain like an unbreakable stone. The tension between class and color is undeniable, heightened even more by the imaginary shattered by this heroic gesture.
I don't know where I read (did I read?) him comparing himself to a black Hercules, faced with one of the many tasks that one has to go through to survive in that inhospitable place. The angle from below to above enhances the scene, whose mythical nature is reinforced by the fact that it takes place on the peak of a mountain, surrounded by other miners, who witness the peculiar event while ending up creating, in this circle, a strange feeling of order and plenitude in the midst of the valley of incessant chaos.
To further appreciate this interesting image, I listen to the recorded comments, available via QR Code, recited by a determined and convincing male voice: “This photograph depicts a moment of tension and confrontation. In the center of the image, a muscular prospector covered in mud and wearing only torn shorts holds a machete firmly while facing a soldier in uniform. The soldier, with a firm posture and a determined expression, holds the blade of the machete with one hand, establishing physical and symbolic contact between the two.”[I]
The description/analysis not only gets the weapon wrong (swapping the rifle for the machete), but also inverts the position between the oppressor and the rebel, giving prominence to the white soldier who holds, “with a firm posture and a determined expression” the blade that threatens him. The error not only misdescribes an image (for visitors with visual impairments) but also imposes, with the legitimacy of an official device, a serious factual and sociological inversion of the analysis offered to the public.
Most likely, it was not the intention of those involved, but the carelessness, haste, laziness and economy in the curatorial and exhibition process not only accepts the risk of such a bizarre error, but also accepts a lack of respect for the work of the photographer, as well as the work of specialists, who should be paid to write this critical/descriptive text well, and also represents a lack of respect for the public, when they are offered digital nonsense legitimized in the form of serious information, in an exhibition of international scope within the cultural space of a large public bank.
Analyzing the image, the error must have been caused by the unusual gesture of the prospector who holds the barrel of the gun with his hand upside down. The non-human reading must have identified this act, by approximation, as that of someone holding the handle of a knife. By doing so, the artificial intelligence (I say this here, because no human would make such a foolish mistake) is not only mistaken in the general description of the scene, which Roland Barthes calls studium, but also in point of the image, an element that contradicts the generalities of the studium, cutting and hurting what is expected.
The cut in this photo that violates the expected situation and, thus, mortifies the gaze of those who look at the image, is precisely the gesture of the miner, which transports the fact of the image to other aspects, fabrications and affections, which the generalizations of Artificial Intelligence will have great difficulty in reaching. This cut did not even need the invented machete, or rather, it is the very fact of him holding the rifle pointed at himself in this way that gives it its strangeness and strength, an unexpected wound in that reality not perceived by the algorithms of technology.
It was precisely by observing that the phenomenology of photography is created in the affection of the gaze that Roland Barthes, without knowing it, already foresaw the limits of the machine in seeing, feeling and thinking about an image: “the affection was what I did not want to reduce; being irreducible, I wanted, I had to reduce the Photo, but would it be possible to retain the affective intentionality, an intent of the object that was immediately permeated with desire, repulsion, nostalgia, euphoria?” (BARTHES, 1984, p. 38).
The lack of affection in a dehumanized gaze leads us to a dangerously smooth and predictable, generalist and idiotic world. The appreciation of Artificial Intelligence may even bring us astute observations of angles and framings, in addition to a correct general context of a banal image, but it does not go beyond the surface of the expected, depriving us of perceptions that open routes for us to travel through deeper and more distant seas, exploring new and unknown horizons.
The intervention of Artificial Intelligence thus broke the uniqueness of the work marked by an experience of specific time and space, which Walter Benjamin calls aura, when analyzing the change in art in times of technical reproducibility, to meet a need of capitalism for greater proximity of the consumer with images. Walter Benjamin observed the destruction of the aura of the image, in the removal of the object from its wrapping and durability, as a change in the form of perception, whose “capacity to capture the 'similar' in the world is so acute that, thanks to reproduction, it can capture it even in the unique phenomenon.” (BENJAMIN, 2008, p. 101).
With the advent of digital technology, this process becomes even more radical, as the matrix itself loses relevance in relation to the copies. However, artificial intelligence has arrived not only to be able to disconnect the image from its original object, but also to disconnect it from the context in which it was conceived and from the perspective of its creator.
A sword and two words
This is reminiscent of the recent controversy caused by a trend of artificial intelligence to transform photographs into drawings in the style of the famous Japanese animation studio Ghibli, emulating, without authorization, the style of its co-founder and most famous director Hayao Miyazaki, known for films such as My Totoro Friend (1988) A Viagem de Chihiro (Oscar for best animated film in 2002) and The Boy and the Heron (Oscar for best animated feature film in 2024).
Famous for still making animation using hand-drawn image by image, each Ghibli film takes years to be finished, sometimes months to obtain a more complex scene, following the strict criteria of its creator. An example of this uncompromising principle was the fight between Hayao Miyazaki and Harvey Weinstein, then head of Miramax who, before being known (and arrested) as a sexual predator, was famous for mutilating films and, in this case, demanded, threatening to sue the Japanese producer, if it did not cut the film Princess Mononoke (1997) from 135 to 90 minutes for distribution in the USA. In response, Hayao Miyazaki sent a samurai sword to the American producer with a card with just two words: “No cuts”. The film was not cut and Hayao Miyazaki would earn respect and awards in the USA and around the world.
Although Studio Ghibli has remained silent on the trend, Hayao Miyazaki had already expressed his view in a 2016 documentary (A series by the Japanese telecommunications company NHK, directed by Kaku Arukawa), when he was invited by entrepreneurs and developers to evaluate an animation made entirely by machines. The answer could not be less categorical: “I am completely disgusted. If you really want to create scary things, you can do that. But I would never want to incorporate this technology into my work. I feel strongly that it is an insult to life itself.”
Even without taking into account the authors' opinions (or paying them), the trend was a huge success. There were already emulators of South Park, Rick and Morty and The Simpson's features, but the repercussion of the Ghibli trend was incomparable, leading ChatGPT, an AI from the company OpenAI, which developed the plugin, to add more than a million users in just one hour. Celebrities and unknowns rushed to have their profiles illustrated with drawings in the style of famous oriental animes. However, in addition to cute portraits, the technology was used for less innocent purposes, such as the cruel stylized meme of a Dominican immigrant crying while handcuffed while being deported, published by the White House's official Instagram profile.
The artistic and humanistic density of a work built over decades was disrespected once again by the empty and unethical speed imposed by those who profit doubly from these technologies: from the number of followers and from the free database they gain to train their machines when a photo is sent to be stylized.
It is no coincidence that in October of last year, more than 11 artists signed a protest against the unauthorized and unpaid use of their creations to feed artificial intelligence platforms. These artists include actors and actresses such as Julianne Moore, James Patterson and Kevin Bacon, as well as musicians such as Thom Yorke (lead singer of Radiohead), Bjorn Ulvaeus (Abba) and Nobel Prize winner for Literature Kazuo Ishiguro. A month earlier, Hollywood stars such as Jane Fonda, Mark Hamill and Pedro Pascal fought for a bill to control Artificial Intelligence, which was vetoed by California Governor Gavin Newsom. Regarding these controversies, OpenAI spokesperson Taya Christianson (in a statement sent to the press) responded: “Our goal is to offer users the greatest possible creative freedom.” Based on the ambiguity between inspiration and copying, technology accumulates data to increasingly develop and profit from other people's creations, destroying not only the aura of the work, but killing the creator himself by alienating him from his creation.
Based on the proposal to increasingly increase the freedom and transparency of information, such technologies insert enormous amounts of information into smooth data flows, without interiority, density, autonomy or dramaturgy of the subject, where everything is visible, available and possible, like a dystopian hyper-truth. This dataism, according to Byung-Chul Han, denaturalizes, through accumulation without depth, the being itself, by distancing it from the knowledge inherent in experience. For him, the use of this information is a form of pornography of wisdom: “It lacks the interiority that characterizes wisdom. […] Wisdom, on the other hand, has a completely different structure. It is tense between past and future. Information, on the other hand, inhabits the smoothed time of undifferentiated present points. It is a time without event or destiny” (HAN, 2015, p.20).
The alienation of the creator by current technologies is, in a way, the death of art as an experience that harms the normality of the gaze, sensitizes people and transforms attitudes, because creation is not just a sample of consumable data, but the result of a free process, at the same time rational and intuitive, opaque, imperfect and, therefore, uncontrollable, arising from the experience of an author in a space and time and conferred by an audience that reconfigures these experiences in their own lives.
Such trends not only disrespect the artistic process, but also affect art at the core of its aesthetic search for something that opens up as a mystery, through a perception that is still blind and unconscious. Byung-Chul Han calls this something natural beauty, which, unlike digital beauty, is completely exempt from conformation for consumption, through the utilitarianism that is integrated into commodities. Artificial Intelligence, on the other hand, disintegrates all the discomfort, all the wound of art to create an immediate and purely positive commodity, in order to establish a permanent self-mirroring, fueled by selfies, trends and likes.
Byung-Chul Han highlights that digital beauty is in a present without history or future, hovering empty of affection and temporality, where the exercise of encounter and dealing with otherness is also lost: “In digital beauty, the negativity of the other has been completely annulled. That is why it is completely smooth. It cannot be torn. Its sign is complacency without negativity, the like. […] Thanks to the total digitalization of the being, a total humanization has been achieved, an absolute subjectivity, in which the human subject is faced only with himself” (Ibid., pp. 40-41).
240 thousand liters of water
As if the copyright issues and the conviction in the aesthetics established over 50 years of artisanal work were not enough, we are also faced with another consequence of the famous trend, which tears at the soul of Ghibli films: the environmental impact. Studies by the University of California, Riverside and the University of Texas (published in the Washington Post) reveal the massive consumption of fresh water to operate such conglomerates of machines behind these artificial creations. Training a single language model, such as GPT-3 for example, requires the evaporation of around 5.4 million liters, taking into account the cooling of the servers and the water used indirectly in the generation of the energy used.
The study also warns that, if this reality does not change, global water consumption caused by artificial intelligence will reach between 4.2 and 6.6 billion cubic meters, equivalent to the supply of 4 to 6 countries like Denmark. This mark would be surpassed in 2028 only by the consumption of data centers within the USA, without even considering the loss of water due to the polluting chip production process.
Following the example of the Ghibli trend, with 6 million posts that it quickly generated on Instagram alone, it is estimated, based on a study by hugging face e Carnegie Mellon University, a consumption of 240 thousand liters of water, which would allow enough for the daily use of water for 12 thousand people (at an estimate of 20 liters per individual). Here we could also add the extravagant energy consumption of these technologies (each image requires the equivalent of charging a cell phone by 25%) and CO2 (Each million images are equivalent to the pollution caused by a typical gasoline-powered car traveling approximately 5 kilometers). It is no coincidence that Adam Altman, CEO of OpenAI, when celebrating the success of the trend, admitted that his GPUs were melting and that he would have to restrict the calculation capacity for a while so as not to crash his system.
Paradoxically, such disregard for the environment could inspire a script for Ghibli's own films, which are heavily anchored in Shinto cosmology, a Japanese spiritual tradition linked to the harvest, the seasons and ancestors, cultivating a relationship of respect for nature, often associated with mythical forces of the world, such as mountains, rivers or the wind. Such harmony is represented in a world where the kami, sacred elements that make up both organic and inorganic things: water, roots, fire and leaves, are scattered. When this peace is broken by misguided human action, the result is always destruction and death: nothing is more current than the increase in natural disasters on the planet due to global warming.
There is no room, however, for an innocent ecology. The nature of these films is at once sublime and mysterious, threatening and voluptuous, generous and deadly. The tension between humanity and nature is at times evident, as in the struggle for survival of a village against the vengeful force of the world, in Princess Mononoke. Sometimes, such a relationship is diluted in the fantasy of the children's universe, as in My Neighbor Totoro, in which the modern post-war world meets traditional cultivated landscapes (satoyama), where the girl Mei learns, from mythical beings, the importance of living in harmony with the universe. Human beings are also capable of extraordinary things, such as the manufacture of airplanes, a visible obsession of Hayao Miyazaki. But even this marvelous machine is often accompanied by the result of human recklessness: death, war, disease.
There is always a path to be taken, a portal to be crossed, an encounter between worlds where survival and understanding always go hand in hand, even if in constant tension. His films portray, with organic lyricism and raw fatalism, the danger to humanity when we forget to respect the universe, such as the post-apocalyptic. Nausicaä of the Valley of the Wind (1984) inspired by the pollution of Minamata Bay.
Hayao Miyazaki definitively demonstrates that there should be no separation between aesthetics, politics and environmental awareness. The cute images of the trend have no right to be so because they disrespect the entire universe of a work: its lines, colors, its contemplative temporality, its poetry and the expression of human relationships and respect for the world. The lack of human and environmental ethics in this event is just a faithful illustration of the decadence of our society.
Incidentally, the Terra Foundation, owned by the Salgado family, which does excellent reforestation work on its land in southern Brazil, could very well prohibit the use of technology that is so voracious of natural resources in its activities. This would be in keeping with the photographer's most recent series, which now invests in recording nature around the world, in works that seek to promote environmental awareness.
Refreshing our perspective on new environmental paradigms would also be insightful in the specific case of Serra Pelada. By renewing our approach with a close eye on environmental issues, we open up another critical layer in which the real culprits are no longer the poor, brutish men portrayed in that mine, but the system in which a few continue to make a lot of money from the exploitation of people and natural resources. As in Hayao Miyazaki, art, politics and environmentalism should always meet in a finality, beyond speeches and beautiful images.
The mythical in the gaze
On the other hand, the negativity of the image, its affections and disaffections, ambiguities, gaps, memories and savagery, make its experience a permanent encounter with the world, with other perspectives, with the dialectical discomfort of the unknown, the unpredictable, the imponderable. In the case of Sebastião Salgado's exhibition, for example, a closer look at these affections can take us beyond the factual and humanist context of his photographs, leading us, for example, to a certain mythical or sacred instance in the photographer's universe. This is what Christian Caujolle observes about Salgado's work: “his photographs send us back to an iconography that is no longer humanitarian, but religious” (CAUJOLLE, 1993, IN: LISSOVSKY, 2008, p.79).
Such teleology makes us understand Salgado's classical framing, based on the golden ratio, as a composition of the scene that refers to biblical or mythological Renaissance themes. Maurício Lissovisky notes that such aesthetics express a revelation linked to the pagan sacrifice of the innocents, being, in a certain way, inscribed before its production: “The latitude of waiting must be wide enough, so that the evolution of the form finds the place that was destined for it from the beginning.” (LISSOVSKY, 2008, p. 79).
In this wait in which the photographer sets his trap to find the record he has preconceived, a determinism imposes itself on the images as in a tragedy by Aeschylus or Sophocles, expressing, through the classic balance of the captured scenes, an unhappy destiny already mapped out, from which the people photographed have no escape.
However, it is precisely this waiting, the time of work and the engaged temporalities, that Artificial Intelligence does not perceive. It does not know of the six years that the photographer, for example, waited for the dictatorial government of Brazil at the time to obtain authorization to access the site and, finally, in 1986, spend 33 days, in the open air, photographing that crater-shaped mine. This trace of existence is perhaps the greatest strength of Salgado's work, often contested for aestheticizing poverty.
However, the photographer undoubtedly captured such powerful images through his presence. It is by participating, as a present body in this universe, that he is able to photograph these destined images. This is the photographic phenomenon that Salgado talks about in relation to his creative process, which, for him, moves away from Cartier-Bresson's decisive moment when the photographic act no longer appears as a single arrow of a Zen archer, but as a siege, which is gradually understood and approached “until reaching the apex of this phenomenon” (Interview with Joaquim Paz In: Paiva, Joaquim. Reflected looks, p.154-5, apoud LISSOVSKY, 2008, p.78).
The experience of this moment makes photography strong through an aesthetic of the irreversible, which, according to François Soulages, starts from existence itself to transcend it, having a continuation after death, at the same time that one cannot return to the previous operation: “This generalized obtaining is not only the image of time, but it is the image of time and the time of the image; its irreversibility is caused by the articulation of time, the nature of the negative and the conditions of its obtaining.” (SOULAGES, 2010, p. 145).
By annihilating the photographic phenomenon (which goes beyond a supposedly captured truth), this error of Artificial Intelligence, which may even seem minimal and innocent, involves us in a cloud of commonplaces, killing not only our intelligence but our capacity to feel and to complement the gaps and ambiguities that make the strength of any image, killing the presence and affection that involves all the agents of the image.
It also ends up killing the apprehension of the time of the image, which would be in Warburg's historical model, as recalled by Didi-Huberman, expressed “through hauntings, 'survivals', remnants, returns of forms.” (DIDI-HUBERMAN, 2002. p. 27-28). Such inconsistencies of time take us far beyond the exchange of two weapons, observing the unlikely gesture of the prospector, as a work reminiscent of the history of the most distant human memories.
It is not just a careless lie, but this lack of interpretative (and affective) density that Artificial Intelligence develops and imposes on current sensibilities. Such superficiality represents the very death of the subject, or rather, of the subjects, who in photography are, for François Soulages (Ibid., p.145), the one who photographs, or is photographed, or looks at the photos, unfailingly linking space and time in the world of the person who perceives them. Before being annihilated, this traveling subject of the images was a subject-becoming, in eternal transformation in the unstoppable composition of the world and, thus, in the construction of his own image: “every time, the subject […] continues to be an enigma to himself: he is unknowable (S = x), above all because he is temporal, and therefore, temporary and irreversible” (Idem).
By making a mistake in the main gesture of photography, Artificial Intelligence distorts facts, violates time and its rituals, and thus breaks with the very principle of photographic irreversibility. The ease and speed that these tools offer not only destroy jobs and create bad students (a key problem for any contemporary teacher) and banal writing, but they also exonerate temporality itself and its nuances, its experiences, discoveries, grief, and learning.
The biggest problem with this technology is not creating simulacra that rival the real thing, because the real thing, as Buddhists have said for millennia, is always an internal construction. The biggest problem is also not the laziness and greed of those who use these means without scruples, even if it is ignoble. The biggest danger is perhaps that of getting lost in a world in which a gesture is devoid of temporality and affection, and a face ends up losing its meaning, whether in front of another face, in a memory, or in a photograph.
*Camilo Soares is a photographer, director of photography and professor of cinema at the Federal University of Pernambuco (UFPE).
Note
[I] Source: https://drive.google.com/file/d/1q67HTOiBgxB24cbYMaYaHhMxUflyfvWP/view
the earth is round there is thanks to our readers and supporters.
Help us keep this idea going.
CONTRIBUTE
