In recent weeks, there has been a lot of discussion about images generated by artificial intelligence (AI). Software such as DALL-E 2, Midjourney, and Stable Diffusion have come under fire for their use. Can a prestigious photography award be given to an image taken without a camera? Is it ethical for an NGO to use images created by this software to condemn police violence? Can AI be trained to use copyrighted photos without permission? These are the questions that professionals ask at the table.
As part of PHotoEspaña 2023 (May 31 to September 3), elDiario.es talks to editor and teacher Gonzalo Golpe (Madrid, 1975). On June 5, La Fábrica publishes When the Wind, one of the first books in the world to use artificial intelligence to generate images. In this pioneering work, the author of the work, Gonzalo Golpe, visually tells a dystopian tale – created with the help of Martin Bollat and DALL-E 2 (AI developed by the company Open AI, the creator of ChatGPT and with the participation of Microsoft) – which invites us to observe our society’s fears and question how we use technology. “AI is not intelligent from a human point of view. Artificial intelligence is not imaginary,” asserts Golpe.
On April 13, for the first time in history, an image created using artificial intelligence won a prestigious photography award, the 2023 Sony World Photography Awards in the category “Creativity”. However, the German artist who submitted it, Boris Eldagsen, lost the $5,000 prize because, as he said in his speech, “AI images and photography should not compete for such an award. They are different. AI is not photography.” Eldagsen performed this performance act to show that these types of competitions are not prepared for the current boom that AI is experiencing. “Not understanding what it means to call these synthetic images ‘photos’ is dangerous,” says Gonzalo Golpe.
Two weeks later, on May 2, the Norwegian branch of Amnesty International published a report on police violence during the 2021 protests in Colombia. While the country’s nationwide strike was captured thanks to photojournalism, the human rights organization accompanied the campaign with several images created by artificial intelligence.
The organization reported this at the bottom of the pictures. However, their use was heavily criticized on social media, prompting Amnesty International to withdraw the synthetic images. According to Amnesty, the decision not to use the photojournalists’ photos in the first instance was made to protect the identities of people who have experienced police brutality.
What’s the point of AI reaching where humans have already reached? The noema of photography—theorized in Roland Barthes’s Camera Lucida (1980)—is: “It was.” In the case of these non-photographs, created by artificial intelligence models that create images from text, the Barthian “it was” makes no sense. “The images they create are the result of a training that lacks critical and moral sense. These programs are heirs and representatives of a visual system based on privilege, discrimination, concealment, and ridicule. That’s why it accepts a person as masculine, young and white”, analyzes Golpe.
The software that creates this type of content is trained and fed on photos taken by humans, which are often copyrighted. That’s why Getty Images, one of the most popular photo agencies, sued Stability IA, the company that owns the open source generative AI Stable Diffusion, which Getty says infringes the intellectual property of millions of copyrighted images. An illegality experienced individually by photography professionals from around the world.
As found by Newtral, in the Spanish framework, the works of at least 55 photographers are illegally feeding LAION, a database that has been training stable diffusion. “I think it is necessary, urgent and even mandatory, to establish legislative controls that require the definition and identification of these images, as well as the control of their use, especially in certain areas,” suggests Golpe.
“These models are educated based on biases, they respond to segregationist interests that seek to isolate differences in order to control them and implement repressive policies. They are tools. Not only human progress, but degeneration and loss of humanity. They were born addicted. Its current development is supported by a vulnerable workforce that is exploited and also disconnected from the agenda of the goals that drive their patents. It is alarming,” warns Gonzalo Golpe.
To quote Susan Sontag, “A photograph in a book is clearly an image of an image.” When the Wind (La Fábrica, 2023) – the first book published in Spain with the help of human image generation artificial intelligence – we are not dealing with a photobook. This visual narrative, through argumentative texts (request) typed by Gonzalo Golpe in DALL-E 2, is the work of the author. A post-apocalyptic science fiction fable with an environmental message. When the wind synesthetic visual discourse is experienced in the context of sadness. An unusual semiotic (and sinister) work with melancholic halos in fluorine tones that characterizes this type of model.
“I use this technology to perform visual composition […] When I was creating, editing and making images, what I was doing was projecting my soul into a kind of visual symphony, as if I were humming the images,” says Golpe.
Gonzalo Golpe is not a photographer. He worked in the image environment all his life. Its appearance and rhythm can be found between the pages of important photobooks signed by artists such as Cristina de Midel, Joan Fontcuberta, Alberto García-Alix and Alessandra Sanguinetti. “I am a mediator between the work and its author, between society and works, and also within the system,” says the editor. Golpe published dozens of texts, the most recent of which were the essays Curso Discurso and Atestado (Cabeza de Chorlito, 2020 and 2021, respectively). Together with Marina Meyer and Sara Arroyo, he is a member of the educational platform Ensambles, which currently manages debate forums on artificial intelligence, and since 2014 he has been part of the graphic collective La Troupe. When the wind is Golpe’s first author’s work. “I come from a poetic background, an obsessive work that mediates between verbal and visual language, so this technology was a kind of epiphany for me. Now I can name what I see in myself and make it materialize visually,” he admits.
“I have always felt the desire to hybridize my voice, to go beyond words and images into more eloquent sounds that can take visual form in the mind. I used to use my own and archival photos to augment my writing, but it seemed too violent when people showed up, which is mostly what I’m interested in. Fiction with people I don’t know, using their images to tell my stories has always felt like violence, a usurpation. At least now I know that the people I am re-creating never existed, and this frees me,” analyzes Golpe.
Golpe has been building an investigation into the origin of the language for several years. This is an essay called Verba Volant, the conclusion of his quest. “I consider myself a remote manager, both as an editor and as a teacher or communicator,” he says.
Photography did not replace painting. The cinema did not replace the theater. Video did not replace cinema. Television did not replace radio. Why do we fear that AI will replace photography?
Etymologically, the word photography comes from the Greek, where “photo” means light and “graphia” means writing. If photography is the art of writing with light, generating an image through text that is fed to AI, what is it? “These are synthetic images that are computationally generated from an algorithm. Nothing they show is real,” Golpe replies. Along with the popularization of artificial intelligence, new terms such as synthography (synthography) and promptography were created, which are still not common in society. Both refer to synthesis and texts (prompts), respectively, that a human must input to generate an AI image. This is a process: a computational calculation based on a proposition.
“From an artistic point of view, I see it as another tool, I don’t think it is necessary to distinguish them from other visual technologies, but when these images are taken as a photograph, that is, as a capture of ‘reality’, they must be used. to be audited by specialists and referred. In any case, fake news is only one of the dishonest uses that happen to me, there are many others and probably more disturbing and harmful”, Gonzalo Golpe solves.
Source: El Diario
