AI is not replacing photographers equally. It is changing the market value of every skill the profession is built on, separating those that were tied to the camera from those that will matter even more.
Ordinary photographs were always replaceable. Labor just hid it. A competent landscape took skill, time, and equipment, and that effort was treated as value even when the image itself was interchangeable. The market paid for the labor, not the photograph.
Then came stock libraries. The value of acquiring an ordinary image dropped sharply: the labor stayed, but the volume of potential sales grew while the price of each purchase collapsed.
AI zeroes out the labor, and the interchangeable image is worth exactly nothing. AI did not take the value away. It made the absence of value visible. And it did this not because it learned to make better images than photographers. It simply stopped counting the cost of producing them as part of the image's worth.
So what survives, it seems, is what cannot be reproduced. Not by the measure of craft, but because it is tied to a single event, to a trace, to an encounter that cannot be generated, because it did not originate in a written concept. This is where photography and generation supposedly part.
The argument holds only while we picture AI as a system that generates images. All of it rests on the photographer's work of hunting for the unique moment. But why can't what cannot be generated be delegated to AI instead?
Not the AI that makes images, but the one that drives the cameras and assesses the situation, the probability of a moment or an event. Or simply records everything at once. If a system can judge the odds of an event better than a person, can take a better vantage point, and can capture continuously, photography's old line of defense starts to fall apart fast.
The need to hold a camera in your hands may shrink sharply in the near future. Not because of generated images, but because of new mobility technologies applied to the same cameras, and most likely because of both at once. Not just to change the angle, but to take the photographer's place in physical space. Today we argue about generating images. In a few years the central question may no longer be generation, but the delegation of the act of shooting itself.
The camera can be handed to the machine entirely. One question remains: what happens to the knowledge a photographer spent decades building? Does it vanish with the camera, or does it just change its field of application?
Go back to the words about a written concept, the prompt. Here is the curious part: precise knowledge of composition and technical language produces far more accurate, more realistic images. Knowledge of the zone system, how spatial layers work, how depth of field and exposure behave, how to place light, and when to release the shutter shapes a generated image far more precisely than mere "vision." A paradox follows. The deeper a person understands photography, the stronger their command of generative systems. AI does not cancel professional knowledge. It moves where that knowledge applies. Apply the technical skills you have used for years to crafting the prompt, the exact description of the result you want, and it will stand head and shoulders above an amateur's. A professional photographer does not describe the picture he wants. He describes the photographic process that has to arrive at it.
Return to the word we started with. A written concept is intention stated in advance. The prompt does not search for the result; it retrieves one that already exists. Shooting works the other way around. Intention and result do not match, because the world intervenes, and the best thing in the frame is often the thing you never planned. These are two different modes: to describe in advance and to discover in the process.
A professional is strong in generation not only because he describes it better. He is strong because all his life he trained something else: to recognize the moment when the result outran the intention. That skill grew out of shooting, out of catching the unplanned. And it carries over into the place where everything is supposedly fixed in advance.
Generation does not level the amateur and the professional. It pushes them further apart. The one without the language writes "a beautiful sunset" and gets exactly what everyone else gets. The one who understands light, foreground, midground, background, and the moment frames a request the amateur simply cannot put together. He simply does not know what there is to say.
Generative images hand us endless combinations. The difficulty appears on the far side of the shot, where you no longer choose the moment but choose the variant that will be final. With the power to spin variations of the same frame forever, when do you stop? When do you make the final choice? The hard part of the profession used to sit before the image. Now it moves, step by step, to after it.
Photographers know a version of this already: pulling the three or five best frames out of a thousand. But it grows far sharper, and demands far more, once endless improvement opens in front of you. Before, the photographer chose among frames that already existed. Now he also has to choose the moment to stop refining. That is a different kind of professional decision.
The photographer who now looks vulnerable from every side is in truth vulnerable in only one narrow place. The same place that always made photography one of the cheapest, lowest-paid trades on the market. That place is the camera operator. The person who exposes correctly. He is no longer needed. But the problem may run deeper. For too long we used the word "photographer" for several different professions at once.
One person searched for the subject, made the aesthetic decisions, ran the camera, edited, selected, printed, and sold the result. We got used to calling this a single craft. It looked single for one reason: all of it lived in one body and was paid with one fee. The client paid for the shoot as a whole and never saw that inside that sum some roles were worth a great deal and others almost nothing.
The fusion was an economic convention, not the nature of the profession. As long as the roles could not be separated, their prices stayed the same. There was simply no way for the difference to show. Running the camera, finding the frame, the selection, and the decision about what the image should be all went on one invoice, though they asked for completely different things and were worth completely different amounts.
AI does what the market never could. It uncouples the roles and, for the first time, prices each one on its own. And the moment it does, the same gesture from the beginning repeats itself. The market once stopped counting the cost of production as part of the image's worth. Now it stops counting the operator's work as part of the photographer's worth. The first role to disappear is the most technical one, the role that passed for the heart of the craft because it needed hands and equipment and so seemed central.
But the curator and the visionary, those invisible layers above the operator's work that always stood behind a successful photographer, go nowhere. If anything, a feel for fine settings, for relationships, for style and composition only sharpens and gains weight. These were always the qualities that set apart photographers whose work could not be reproduced by simply repeating the technical process.
What is striking is that AI takes away the chance to make a good photograph by accident and become known for it. The chance we still see at global photo contests, where the winners are excellent works by authors not yet twenty. That stroke of luck disappears once generative technology and camera control change radically.
Maybe that chance is what we are really mourning. If so, the argument about AI turns out to be no argument about technology at all. It becomes an argument about which part of the profession was ever truly ours, and which part only needed human hands for a while.
AI raises the stakes. It asks for far more professional competence than many who call themselves photographers actually have.
Join the Fstoppers community for free
-
Post comments and join in the discussions
-
Browse the site ad-free
-
Share your work and get featured in the community
-
Compete in the photo contests for fun and prizes
36 Comments
Hi Alvin, this was an interesting read but i disagree on several points. Though I wonder if that's intentional?
The first thing that seems very off is to suggest that all worth stems from a flat monetary value. This is completely misrepresentative, there are all sorts of factors to consider there that might alter the monetary value that someone will pay a professional to make photographs of something.
The second point is more worrying to me, that a person behind a keyboard is anything close to a photographer, and that photographers will make better prompts. It might be technically correct in the soulless, growth above all world the tech companies seem to want us to live in but it only works if you consider only money. As a lot of your points seem to be geared towards commercial photography, how about if I want my products to not just look authentic but BE authentic. If my brand is built on authenticity (quality of manufacture, etc.) I'm not going to risk cheapening that image using some generated image that gets the colour or the details wrong, I'll do it myself however I can or pay a real professional. And to anyone willing to say you can get it with the right prompt, I say maybe, but the time spent checking, refining, rechecking, refining, rechecking again over the whole image will mean the costs go up if you use AI, not down. It's happening with AI generated training videos already.
The article also treats a photograph as the same as a generated image. It is quite demonstrably not. The worst photo made by a person with a camera required them to be there. It's long been conventional wisdom that the harder something is to achieve the more worthwhile it is. Good things come to those who wait. An image generated of a soon to be extinct species by AI, worthless. The photo from the last member of the species taken will be priceless (morbidly I know) precisely because of what it is, moreso if it's taken on film and there's only one!
I think we should call Generative AI out for what it really is. An attempt by the rich tech companies to devalue human art and technical expertise, as a way of keeping working classes down and removing social mobility with out compliance. As goes for photographers, the same will happen for musicians and other artists making digital art. Physical art might just survive because that's another thing entirely.
Thanks for making me think so hard :D
Thank you, Ed. I agree that the commercial value of authenticity is the main thing. For documentary, artistic, and reputational content, the origin of an image can indeed be part of the product. With commercial photography, it is not so clear-cut.
And I am not calling the person behind the keyboard a photographer. I am saying that photographic knowledge transfers to a new field of application, which, in my view, is a different statement.
As far as the time to produce an image using AI software vs. shooting the photo myself, I'm leaning toward AI being tremendously faster, no matter how many revisions. For a straight picture shot from my front doorstep, maybe not, but AI holds the advantage with about everything else.
Years ago, before everyone had heard of AI, I attempted to recast my San Francisco cable car picture into a Victorian era scene, using Photoshop to delete present-day cars and people, and replace them with characters I had sourced from the US Library of Congress. It took me three days to get to my finished picture.
I chose the same original picture a week or two ago as an experiment for what AI could do, and with minimal inputs, it produced the image in my other comment in about three minutes.
Wow, that is a lot of blabla. AI generated imagery has nothing to do with photography, whether if it's used by an amateur or a pro. AI generated stuff can't be called a photo. It can be art, an image, call it what you like, but not a photo. Question for the writer: what do you call an image a photojournalist makes, who tries to do justice to reality? A 'real photo'? Or even fake news maybe like some pigs in this world? Please draw a line somewhere: my line is generated stuff, and yours Alvin?
I think generative imagery is not photography in the documentary sense. But we should consider the question not only from the standpoint of the person who makes the photograph, but also from that of the person looking at the image. For the viewer, its generative origin is often not apparent. The point is not that the media are identical, but that some of the professional knowledge developed through photography retains or increases its market value outside the act of recording reality.
I am not against generative art, even if made by someone with professional photography knowledge. It can have value, if valued by someone watching the result. This does not make it a photograph, even if the observer can't tell if it is or is not.
I think generative imagery is not photography...period. You add 'in the documentary sense'. Generated art is just that, just like a painting is not photography either. A painting made by someone with knowledge of photography is not a photograph, even if an observer can't tell if it is a photo or a painting.
I can only imagine you yourself want to be a photographer, and you want to use generative AI. But please do not call your generative imagery a photo. Is it so bad to be called an artist instead?
I think you’ve misunderstood my position. I don’t work with generative images at all. Everything I make is created with a camera, including my studio portraits, still lifes, and the occasional street photograph. At the same time, I don’t think of myself primarily as a photographer. The camera is simply the medium I’ve consciously chosen, much as a painter chooses brushes.
The question that interests me is different: what remains uniquely the author’s once a convincing visual image can be created without a camera, and the viewer can no longer reliably tell the difference? My answer isn’t “the photograph.” It’s the author’s gesture. That’s the one thing that still cannot be generated in any meaningful sense.
The time appears to be coming when the machine will not only generate the answers, but ask the questions. And then what do you think of human gestures? Are you sure they will continue to be unique?
A world dominated by machines over humans who are helpless to intervene is arguably everyone's worst nightmare, but that appears to be the direction we (AI) are heading.
I'll be honest: your language soup wears me out. Just to be clear, you wrote 'I think generative imagery is not photography in the documentary sense.' In your reply you manage to ignore my remarks about the 'in the documentary sense'. So again: where do you stand? Can generative imagery be a photo in a non documentary sense??
Speaking of tiresome....
I would really like to know why the opinion of Alvin, Patrick, myself, or anyone else matters to you. You've made it clear exactly where you stand, several times here and in other articles... AI both in whole and in part is not photography. You've made up your mind, and it appears there's absolutely nothing anyone could say that would change your mind in the slightest. So what's the point of this conversation for you?
I do not care if I wear you out Ed. Your endless looooong comments do not really bring me anything, but untill now I did not complain.
I asked Alvin a question, not you. If you read carefully he doged a remark I made. This explains my repeated question. I respect anyone's opinion, but yours a bit less from now on.
No, I would not call a purely generated image a photograph. By “in the documentary sense,” I meant photography’s function of recording something that was present, not documentary photography as a genre. But a generated image does not need to become a photograph to replace some of its commercial uses.
And partially generated? Nah, nevermind. I do not want to keep on torturing my friend Ed.
Commerce, sure. Sadly the awarded images on this site have lost all value to me. I just cannot tell what is AI and what is not. Sad, a hobby going down the drain. Even worse: photography in general losses credibility. Fake news, 'nudification' tools, extortion....well, hurrah: AI made advertising much cheaper 😬
If word soup and long uninvited comments bother you, maybe you're in the wrong place.
No Ed! I just love reading 8 looooong comments from you on every single article on this site, because like you I have nothing better to do like taking actual photo's. 😬
Okay... I will make this as short as possible, so as to not tax your brain. You know virtually nothing about me, nor do you seem to care. All you seem to care about is driving home the point that AI is not photography. Fine. I get the message. Having nothing better to do may be true for you... you show nothing of your photos here. Do you actually make photos, or just criticize the work of others?
However, if you'd bother to look at my Fstoppers profile, you would clearly be able to (without thinking too hard) understand from seeing the pictures (no reading required) that I do indeed have better things to do, or can at least do more than just engage in silly arguments. I do, indeed, spend the better part of my time making photos. I can also elevate sarcasm to an art form, just as you can.
It's tempting to comment on your images, but there is no way to tell if any of them are real. That is exactly the reason why you can't see my photo's: it would be impossible for you to know if they are generated or not. Correction: I have not criticised anyone's work specifically, I've only made general remarks.
My conclusion: I think you are tiresome, you think I am tiresome. Let's not read each others comments. And please do not anwer questions not intended for you. Thanks Ed 👍
Lest you forget... this whole topic of conversation began in the AI contest and you asked my opinion. If only you knew where the discussion was headed, you undoubtedly would never have asked. But remember, too, that you were frustrated because Patrick wasn't exactly answering your questions in the manner you expected. I volunteered an opinion. If you want a private conversation limited to yourself and the author, I suggest using the private message feature. Otherwise, your comments and questions are fair game for the entire community. That's how this place works, so I can not honor your last request. Your best plan is to not read comments written by people you do not like. And I can make my own decisions in that regard.
Amai Ed, you are distorting reality now. Maybe take a deep breath and relax?
'Never have asked'?? Wrong. Patrick not answering a direct question: pathetic. Not frustrating, it was an answer in itself. You tried dodging the same question but finally answered: 'fully AI generated is not a photo', you stated. Again: I have serious doubts looking at the images you placed on this site supposedly being about photography.
Sure you can comment on my comments, but please do not answer questions not directed at you. Thanks again Ed 👍
Okay, I'm relaxed now. Now to refresh your memory...
You've raised your same question about AI recently in several articles. Remember the one about Ansel Adams? Pop... you asked me specifically, in your words: "I am really curious where people draw a line: if a 100% AI generated picture is not a photo, what is? Or do you think 100% AI can produce something that can be called a photo Ed? Patrick Hall did not answer this question, I hope you are brave enough to reply 😁"
So my reward for being brave and replying to you is the accusation of distorting reality, and having my photos debased. Thanks a lot. Don't tell me that your question was not directed at me. You are quite mistaken.
Proof is in the screenshot attached below. The person named Ed is me, followed by a question mark, and "people" seems like an open invitation to other people to comment and answer your question as well. You may have the last word. I am done here. My apologies to Alvin for hijacking his article over this senseless argument.
You lost it Ed, so yes I must reply again. I was not referring to the proof you show. I was referring to you answering to a comment i wrote on this article not directed at you.
And distorting reality is about your comment that 'If only you knew where the discussion was headed, you undoubtedly would never have asked.' And then you wrote I was frustrated. Both statements by you were poo from a bull, distorting reality. As you have seen I started the same discussion again 😬
I hope this gets your head straight again Ed. Debased? Mwah, I do not think calling out your presumed use of AI should be called that. You are not a victim Ed, grow up please.
This discussion shows why photographic theory and education (so-called word salad) still matter. The question Pop keeps asking has no real substance, because its answer was settled long before AI arrived. It rests on a particular property of photography, its indexicality, the trace of an event without which a photograph cannot exist. The concept of the index comes from Charles Peirce, and Rosalind Krauss applied it to photography as early as 1977. Roland Barthes described the same intuition in other words, through the "that-has-been."
I am to dumb to understand your writing. What is the relationship exactly between the semiotic theory of Charles Sanders Peirce, and the use of generative AI, or a pursuit of truthfulness if you wish, in photography? Roland Barthes later in his life said farewell to the semiotic theory did he not? His 'that-has-been' did not describe the same intuition as you write?
I am a simple person: for me the use of generative AI/ photoshopping is fine if you call the result art. But not 'photo'.
Or in practice: it would be quit something if we discover the photo's of tank guy on Tiananmen Square and of Phan Thị Kim Phúc were shopped.
Maybe you can explain to an uneducated soul like me what substance I am missing?
The connection is direct, and you spell it out yourself. Your Tiananmen example is indexicality. It matters because it happened in front of the lens, which is exactly what a generated image lacks.
Indeed, Roland Barthes left indexicality behind. Was this a hickup by you using AI? 😁
Well, lovely. The real problem off course is that AI generated imagery will be/is indisguisable from true photo's. Indexicality can't be proven with 100% certainty. And yes, that is a problem to me. That is the substance my questioning seems to lack according to you? I wonder how historians will look at our imagerysoup in 100 years. I hope they'll have the AI to disguise AI from non AI 👍
"AI Is Repricing a Photographer’s Skills": yes, lets celebrate this sign of human evolution.
If you dig so deep, why all these questions whose answers were settled long ago? (This is a rhetorical question).
That sounds not so inviting, but probably to your delight an anwer anyway. Nothing has been settled. If you can do a mortal soul a favor, and come down from the mountaintop of wisdom: help! Demand regulation, fight. Grok as the future truth? Are you not worried at all?Or just be content with yourself up there, it's up to you 👍
You said in a post-article comment, Alvin: “For the viewer, its generative origin is often not apparent. The point is not that the media are identical, but that some of the professional knowledge developed through photography retains or increases its market value outside the act of recording reality.” I agree with your first point; unconvinced about the second.
That would be nice, but I’m not sure the argument in favor of a photographer’s skills correlating with value holds up. It seems to me that about anyone with just a simple idea of what they want can get amazing results from an AI image generator. Commercial art consulting firms tasked with sourcing hotel art are adopting AI faster than you can imagine. It may not be a photograph in the traditional definition of the word, but that rarely matters to an interior designer tasked with filling a bank or hospital with decorative art.
On a practical level, the theory of what a product or service is worth goes down the drain the minute a prospective customer calls and asks if you can solve an urgent problem for them. They might be willing to spend a lot of cash to have me produce a simple AI image, even when the time necessary to make the product or provide the service is minimal, or requires nothing very complicated. They may not have the time or desire to learn how to produce so much as an ordinary picture with the most basic AI text inputs. And then the question gets more complicated: what to do with the digital file after having made it? Most people have no idea how digital image resolution or monitor luminosity translates to making a canvas ready to hang on the wall. Or where to go to have it printed. It's kind of like when Microsoft Publisher hit the streets and graphic designers thought that it would kill the need for professional services, since the boss could simply assign the task to an office manager. But even though they could design their own brochures, many businesses didn't. And if they did, the print files handed over to a printer were often a real mess. Image creation doesn’t work in a vacuum. People who buy or make digital image files have to figure out what to do with them. And that’s where I’ve historically solved problems and created value for my customers. To your point: accounting skills don't easily translate to graphic design and printing skills.
You asked: What happens to the knowledge a photographer spent decades building? Good question… I’m glad I’m old enough to not have to take the answer too seriously. The question I have now is whether or not to embrace AI as a service to complement my photography. If I were 21 instead of 71, the answer would be definitely yes, I would be on top of AI in a heartbeat. At 71, it boils down to the direction I feel is most gratifying, without having to worry about a mortgage on the house. I can stick with the known parameters of traditional photography, or adapt to the future. And as surely as the automobile replaced the horse for the way people travel, AI tools will render the camera and photographer's skills less and less important. AI image generating software is kind of fun. It can take me to a time and place that I could not go otherwise. The work is not hard and it could solve a problem for my customer. Those are real world implications. The pictures aren’t going into museums or galleries, so the question of whether it’s made with a camera or AI is generally irrelevant. A photographer’s perspective on AI image creation would seem to produce better results, but I hesitate claiming that it’s fundamental to the value I offer my clients. Either way, it’s uncharted territory for me.
I actually agree with almost everything you’ve written.
My point wasn’t that photography automatically gives someone an advantage in AI image generation. As your example shows, people without decades of photographic experience can often produce remarkably good images with very little guidance.
What I was trying to describe is something slightly broader. Photography has always involved much more than operating a camera. Understanding a client’s problem, making aesthetic judgments, preparing files for production, and delivering something that works in the real world are all forms of visual expertise that don’t disappear simply because the image itself may now be generated rather than recorded.
I also think your Microsoft Publisher analogy is an excellent one. New tools rarely eliminate professional value; they tend to shift where that value is created.
The question that has become most interesting to me is a different one: if a convincing visual image can now be created without a camera, what remains uniquely the author’s? That’s the question I’m trying to explore, and I suspect the answer lies less in the recording process itself than in the author’s gesture and the decisions behind the work.
I think you've hit the nail on the head in this and in the thread responding to my comment Alvin. The experience, the process, and the result are all the author's to charge for in my mind, as they contribute to the reality behind the image. The more exclusive the process (e.g. large format cameras few have or know how to use well), the better the experience (such as how easy or pleasant the person is to work with) and the more reliable and high quality the results (either in trusting you'd get the quality required or the results are delivered in the best fashion and time frames as examples), the more they can charge. Someone might not want to pay me for standard, white box product shots any more, but that was likely a very small or dying market for the reasons you've noted in the article. AI images should always essentially be worthless and any business paying someone else for AI generated images is kidding themselves.
I hope for the sake of knowing and valuing what's real that people keep putting value to things that take real, lived experience and time, because if they stop then we might as well plug into the matrix and play in VR while the world burns.
Thank you, Ed. But here it is important not to conflate three things that I want to highlight:
1. The absence of an authentic photographic event—this matters for the documentary function of photography; it is the value of the moment and the value of being present.
2. The absence of costs or a unique craft process—how much labor has been invested in producing the final result, especially when we are talking not about a digital photograph but about a physical object.
3. The absence of market value in the result—for example, when a photographer delivers the result more slowly and at a higher cost while its value to the business remains the same.
First of all, I attempt to frame my comments less in a manner of agreeing or disagreeing, and more on simply elaborating on a thought. It may sound like an argument to you, but (most of the times) is not. Most articles that I respond to just inspire another thought, maybe or maybe not precisely on topic of the article. Which also means that I understand more of what you were trying to say than you think by my comments.
That said, I'm confused by your comment: "My point wasn’t that photography automatically gives someone an advantage in AI image generation."
Because in your article, you said this: "The deeper a person understands photography, the stronger their command of generative systems. AI does not cancel professional knowledge. It moves where that knowledge applies. Apply the technical skills you have used for years to crafting the prompt, the exact description of the result you want, and it will stand head and shoulders above an amateur's."
It sounds to me from your article that a photographer's skills should indeed produce better AI imaging results. When discussing customer problems such as printing, you and I are saying exactly the same things.
The question you pose: "If a convincing visual image can now be created without a camera, what remains uniquely the author’s?" My answer: probably a lot less, maybe nothing. But how much does it matter? That's for each person to answer for himself and whether it's a personal decision or a business decision. My type of customers are filling a need to decorate a wall. Authorship doesn't matter. For a museum or gallery, I'm sure it means everything. For an artist deeply connected between intention and results down to the pixel or brush stroke level, the process matters greatly. I will say though, from my first experience with AI generating the cable car picture, that I felt more an integral creative part of the picture than I had anticipated before launching the project. It does feel like my hand is in the picture. In other words, I would not hesitate signing my name on the print.
Unique is a different story, as the same image could be made by another person, but that is generally true of a photo made with a camera as well. I know you've tackled that subject within the confines of photographic art. I suspect I see less opportunity to create something entirely unique than you do, no matter which tools are used.
I think I didn’t phrase it quite correctly. What I meant was the technical skills involved in using a camera. But beyond those, there is also the photographic eye, vision, and a sense of color and composition. And those haven’t disappeared. What matters is not how you create, but what you create. In my view.
Okay... thanks for the clarification. That makes sense. I think my vision can manifest in an AI generated image, as my examples here would prove. But can a machine have vision of its own? Can it define its own purpose? Can it feel anything? Then what are my skills worth? That seems to be the stage where the most human of creative skills could be replaced, and the direction where AI is headed. That is the scary part to me.
These are questions about the future, and any answer to them is a guess for now. But they sit on the other side of meaning. The question is not what the machine will do, but how a person responds to it. A visual image always appears inside a context, and its importance is set by that context, not by the image alone. Can a machine-made image be worth more than a human one? Yes, without question. Can human authorship be irreplaceable? Yes, without question. Both are true, because they answer different situations, not the same one.