Recent images like those of the Pope Francis in a white down coat wave of Donald Trump violently arrested by a group of policemen have triggered the alarms about Artificial Intelligence (AI), capable of producing incredible hyper-realistic images. The algorithm and its application to the world of the image generate a multitude of questions. For now, a series of recurring failures -especially in the detail of the hands and their position- helps to identify artificially created photos from those captured ‘in situ’ by a camera. What will happen the day that technology improves and we can’t distinguish between true and false images? It is not only a challenge for society and democracy, but also for artists who make a living from design, photography, video, film and special effects. Will they, too, be replaced by technology?
David Holz, CEO of Midjourney, has already announced changes after the controversies over the photos of Trump and the Pope. From now on, the bot that facilitated the creation of images will no longer be free and has been implemented a monthly subscription ranging from 10 to 60 dollars.
Edward Hopper or Leibovitz style
Image generators have not stopped evolving and with instruments such as DALL-E 2, Midjourney and Stable Diffusion Anyone can get a dream image in a matter of seconds. It is enough to specify what you are looking for with the words ‘et voilà’: it appears in less than a second. You can even specify what type of lens we want to focus on, the plane we are looking for, the model, the background, the type of camera, film and photographic style. You can choose between 1,000 entries with names of artists of all times and styles, from Edward Hopper to William Eggleston or Annie Leibovitz.
This type of technology is generating a lot of fear, especially for those who work for image banks. “There is a very beastly disruption in the market”, says the photographer Ferran Izquierdo. There is a paradigm shift and you have to adapt to new times. Izquierdo is head of studies at the Idep School of Fashion, Graphic Design and Photography, where this year they have begun to teach AI classes to their students. “Image generators are one more tool in the creative process,” he says. “The more you know about photography and aesthetic fundamentals, the better you will be able to use new technologies. But it’s a living world. Maybe in three months everything has evolved.”
‘Software’ focused on porn, manga and actresses’ faces
Franc Aleu, collaborator of La Fura, is very hooked on AI. He was one of the first to use Midjourney and now experiment with free software emerged after the release of Stable Diffusion. “The ban has been opened because it generates very different models of images. There are focused on porn, the ‘deepfake’, the magician, a type of painting, the faces of actresses… there are thousands“. Aleu is a descendant of a saga of photographers and confesses that he loves to feel part of this new world that, just as happens with social networks, already knows censorship. “In Midjourney if you write at the prompt: “sin of gluttony” or “tit” nothing comes out. Sin, gluttony and tit are censored. But if you put “boy exploding with grenades” no problem.”
“All my life I have said that images are always illusory, that photography is an interpretation of reality and not a reflection of it. Now with the appearance of AI it is obvious”, says Joan Fontcuberta
More than 2 million images per day
This year, more computer images have been generated than all that have been made throughout humanity. According to ‘The New York Times’, in October DALL-E 2, the image generator that OpenAI launched in the spring, already had more than 1.5 million users creating more than 2 million images every day, according to the company. Midjourney, another popular AI image generator, has over 3 million users on its official Discord server. Google and Meta have also created their own image generators, but have not released them to the public.
live with doubt
“All my life I have said that images are always illusory, that photography is an interpretation of reality and not a reflection of it. Now with the appearance of AI it is obvious”, points out Joan Fontcuberta, a reference in Spanish photography. The credibility of the doctored images of the Pope and Trump are a sample of what is to come, a society that “must live with doubt as long as there are no ways to find certainty”. We are facing a change of era, discovering a new world. “The change of visual regime is as important as when the photograph appeared. At that time, images that were not made by hand had never been seen. Imprisoning a gesture in a daguerreotype was something that could not be understood but it was fascinating. Now the same thing happens, we still do not understand what Artificial Intelligence does but we appreciate it because the result is disconcerting and wonderful. We are taking the first steps, we still haven’t mastered the medium.” For Fontcuberta, the most interesting thing about AI is exploring what traditional photography does not have: “the latent space produced by the accidents of technology, the imperfections it generates.”
“This hooks. I have become obsessed
Related news
The same is the opinion of Ariadna Arnés, who after 30 years dedicated to photography is discovering a new world with AI tools. Two months ago she started working with AI and two of her creations were chosen for exhibitions in two New York galleries, organized by Obscura and Joyn. “I haven’t sold them yet. The NFT world is very inbred. Collectors go to names that are already established, looking for profitability rather than discovering potential,” she explains. Arnés admits that it feels like when she started as a photographer. “This hooks me. I’ve become obsessed, is that it is amazing to capture everything that comes out of my head without leaving home. Enough with the computer.” For her, it doesn’t make sense to replicate conventional photography, but the possibilities of “playing with surrealism, with strange things like people with four legs result of system error”.
Strange and too perfect faces
Can AI replace the photographer? For now it seems difficult because in most cases it is easy to identify the images generated with AI. On the one hand, the beauty patterns he uses are unreal: there is something strange, too perfect in the faces it reproduces. On the other, it is enough to look at details such as the hands or teeth of the models. But technology advances very fast, new things emerge every day. Still life, illustrative photography or for image banks such as Photostock will soon disappear with the advent of this virtual world. “Detecting the deception between the real thing and the artificially generated images will be increasingly difficult. Even I, a photographer, have a hard time telling the difference sometimes“, admits Arnés. And Aleu remembers that if the electronic book has not finished with the physical, traditional photography will continue to exist, adapting to the new times because there is no going back. “Now you can do fashion campaigns or find any image from the computer. Many trades will go to hell. It’s scary. If I was young I wouldn’t know what to do“.