But after all, what is Nature, but one great field of wonders past our comprehension?
H.F. Talbot
We can never understand a picture unless we grasp the ways in which it shows what cannot be seen
W.J.T. Mitchell
One photographer, Miles Astray, won two awards (“1839 Awards”, bronze and People’s Choice) for a photograph submitted in the AI (artificial intelligence) category. Here it is! The photograph wasn’t generated by AI… It’s real.
So if it’s not an AI photograph, there’s still something wrong. What happened to the neck? I have to admit that I thought Astray had “erased” the elements in question with a software. And then I looked at a number of flamingo photos on the Web, and never in my life did I think I’d ever looked at so many flamingo images, since, if I’m very fond of certain birds, I’m not at all a flamingo fanatic. I couldn’t care less about them in general. I don’t even think I can stand them. They’re too gregarious for my taste. Anyway, the point is that, even after reading the author’s words reported below, I really thought that there was something fishy going on. And then, a few hours later, back from a walk with a 9 scale wind in the face, I look again and tell myself that the flamingo has kept its neck and head, and that they’re just on the left side. See the little protuberance on the lower left? That’s it! The birth of the neck. And indeed, the flamingo scratches its side. There’s no doubt about it, Phoenicopterus roseus is taken from the front, as you can see from its webbed knees and feet. That’s how far I’d got by 2.50pm, before going out, and in the meantime, having contacted the artist, he got back to me, at 7.17pm French time! To his enthusiastic reply, he was kind enough to attach a montage:
Thanks to Miles Astray’s indications, we can see that the base of the neck is just at the crotch (but does a pink flamingo have a crotch?), while its beak is where indicated. Photographed like this, frankly, it almost looks like a gag; then again, flamingo specialists must not be a dime a dozen, and I beg to ask: Did you know, before reading this article, that such a bird could make its neck and head disappear in such a fashion? Well, now let’s ask another question. Astray submitted his photograph to the prestigious “1839 Awards”, in the Photography AI Category. And so he won, and two awards on top of that! And then just afterwards, Astray contacted the Jury, and spilled the beans, as he indicates to me:
We don’t know how long this hope will last, but what we can assume is that the day will certainly come when no one will be able to distinguish between native photography and AI images. But we’re not there yet (though?). Let me now turn to an attempt to reflect on this episode. Astray submitted this photograph for the “1839 Awards AI”. It was a joke, as we all know. Except that it worked! But the question, then, is What could possibly have prompted the Jury to award the “AI Prize” to the “Flamingone” photograph? It seems that none of the Jury members had ever seen a flamingo in this position (are they’re to be blamed?). If such an evidence had appearead, then the deception would have been immediately discovered and the photograph disqualified on the spot. So it’s fair to say that the very ethology of flamingos poses a threat (for the time being) to AI imaging! And one thinks of our exergue, written by the great and admirable Henry Fox Talbot:
But after all, what is Nature, but one great field of wonders past our comprehension?
Because it’s quite likely that it would be perfectly possible to produce thousands more natural photographs while at the same time injecting a good dose of indecision, and therefore doubt. The question that also naturally arises is, in what way could Astray’s photograph even be identified as a product of AI? This begs polymorphous questions, linked to (1) perception, (2) the comprehensive aesthetics of images, and (3) philosophy.
(1) From the naive point of view of perception, we first see a pink flamingo. Then we wonder where the neck and head have gone. We look for images on the Internet, even entering “flamingo hiding its neck and head” in the search bar, but we can’t find anything equivalent to the Astray photo. Returning to the image, we can assume that Astray has erased the head and neck, or even modified the body of the pink flamingo, since image searches have revealed that the bird’s body is ovoid rather than spherical. Suspicion therefore grew. But we’re caught in a doubt, so we try to get to the bottom of it, but it’s confirmed that there’s nothing modified here. (2) The flamingo’s position is quite astonishing, when you’re not used to studying them… As we’ve seen, it leads to a certain seriality of questions. Reflecting on this image, I come to that native expression, “comprehensive image aesthetics”. What’s interesting about Astray’s photograph is its almost surreal aspect — a bird without a neck or a head — and yet standing upright. It was, I suppose, this surreality that must have struck the Jury and enabled them to award the AI prize. But now (3), I wonder what led the same Jury to think for a second that this was an AI-generated photograph? Because, factually, there’s nothing extraordinary about seeing a flamingo on a beach… And while the jury may perhaps have been astonished by the enigmatic disappearance of the neck and head, it wasn’t even necessary to use AI to make them disappear — software such as Photoshop or Lightroom could do very easily the job. The question that may arise, in fact, is the one that, faced with a perception that is nonetheless very real, that exists in reality, appears to produce in the viewer’s eye a lack of understanding that leads him or her to judge, precisely, that such a reality does not exist. We’d like to ask the Jury how they could have seen an image so obviously created by AI. Since we are unable to know about this, we can only speculate. It’s astonishing that the Jury went straight from thinking of photography as real, to AI’s invention of an image impossible in reality. It’s almost like forgetting the history of photography, which includes, almost genetically, a kind of DNA of manipulation. For example:
Mumler knew perfectly well that he wasn’t showing spirits, since he was the one who double-exposed his pictures. But his customers believed him. He advertised on that basis. He would sit the customer down and ask him to close his eyes and urged him to think very hard about the deceased. This concentration was bound to summon the spirit in a ghostly fashion! Needless to say, this kind of set-up flourished in all countries where photography was used, leading to the emergence of “spiritual photographers”, notably the British William Hope:
See, there’s no artificial intelligence here, just a simple, hand-crafted photographic manipulation. On the other hand, from both Time and History, we still have doubts about what we’re seeing, as evidenced again by what happened with Boris Eldagsen’s photograph, which won the “2023 Sony World Photography Awards” (SWPA). But Eldagsen turned down the prize, revealing that it was an image invented from scratch using the DALL-E Internet platform.
Eldagsen’s aim was to engage in a critical dialogue between “real” and “fake” images. In a final statement, the “2023 Sony World Photography Awards” Committee claimed to have known from the outset that Eldagsen’s photograph was “fake”. But then, if they knew this, why did they award it? It’s rather curious. As we can see, in the early days of photography, the question of truth was posed — one might even say, as early as the “Point de vue du Gras”, the world’s first photograph, whose exposure time took six hours (we talk about it here). Intrinsically, it seems, the “life” of photography, from its birth, had to accommodate itself to reality, and thus the truth of reality. And that’s exactly what we’re blaming on an image that, supposedly rendering reality, can either mimic or illustrate it, or even surpass it. The problem could be solved by saying that there is only one reality; and that any image of reality, whether mental and/or pictorial, represents either a “copy” of reality, an interpretation of reality, or a purely hapaxic creation. In other words, those who believed from the outset that photography was faithful to reality were mistaken, for while a photograph can indeed bear witness to the truth of reality at a given moment, reality is not as flat as a Hahnemühle sheet.
The amazing thing about Eldagsen’s image is that you can see strange flaws in the main character’s pupils, or in the hand resting on his left shoulder. Look at those fingers. It’s all wrong. Similarly, the hand on the right shoulder is huge, and look at that forearm, which looks like an aluminum tube, and is out of alignment with the wrist! And the left hand on the right breast is far too “perfect” (it too seems to be made of an alloy). On the other hand, our flamingo is not alone. In a more sophisticated, but kitschier register, we can find this image by Suzy Dougherty:
She entered a photo competition, with the image above. Her proposal was rejected, on the pretext of a suspicion that the image had been produced by AI. However, according to the photographer, this is not the case.
In the end, the best photograph in this assortment is that of our good old pink flamingo (the first pink flamingo fossils date from the Middle Miocene, some 15 million years ago). With two times nothing — a neck and a head that seem to have disappeared — the image is both a joke and, at the same time, an enigma: Where has the Phoenixcopterus gone?
Leon Myshkin
contact: mychkine@orange.fr