I hear you, and thanks for the note. I worried this might put some people off — and for good reason. I was inspired by the ideas and wanted to share them, so I figured I'd take a risk and put this out there. Thanks, hw.
I am not sure AI is so different from us. Most people default to generic answers unless they are encouraged to go further. If you ask a shallow question, you receive a shallow response, whether from AI or a person. However, when you challenge it, reformulate, and question the question itself, the discussion deepens.
I see this frequently in teaching. I have 56 students, and most of them initially repeat generic ideas. What they have read, what they have heard, what seems logical to say to avoid sounding foolish, or what someone in authority has already stated. Real thinking begins when you ask, “Is this actually my view? What can I add from my own experience or observations?”
In that sense, AI feels more like a mirror. The depth depends on how willing we are to think, and how much we are prepared to push it, or ourselves, further.
Hey Nathan, there are plenty of reasons to take issue with AI. I certainly do for many reasons. It's also probably in our best interest to not be naive about the capabilities of these technologies. Claude has managed to answer a photo 101 paradox using resources most photographers will never read. This post and the answer Claude provided is a philosophical question in itself. Considering substack was started as a platform for writing, it's probably in our best interest to know that AI can answer a foundational photography issue better than most photographers. Personally, I would like to ignore AI all together. I think it's likely that the broligarchy is going to try to use it to take our money and attention away. It's going to be making perfect realistic pictures too. We are crossing a threshold and without understanding what it is capable of we can't begin to know where or how to move forward. I wish this had not happened during my lifetime, but I cannot deny that Claude’s response here seems to address this very issue in the last sentence; “The question “Does the camera lie?” might ultimately be less interesting than the question it opens up: What do we want from truth, and why did we ever think a machine could give it to us?”
Hey, Randall, these are all excellent points. And I agree with that last statement. I find myself asking Claude questions from time to time, mainly to inspire reflection and to help me work through my thoughts. I had a feeling that doing this in public might catalyze some community conversations. I hope it does.
I believe tech companies are coming for everything we do, such that in the future there will hardly be room for human interaction that isn't filtered through the distorting lens of AI. And so I am currently operating with the theory that every inch I give to the LLMs is space taken away, exponentially, from humans. And I want to engage with people thinking and saying and doing human things. So — successful provocation aside — it's a no from me, dawg.
You know I'm a fan, Bill, so I appreciate your feedback. Fear not, this won't be a recurring feature! Still, the ideas are compelling, and I know you appreciate them. I learned something new from these philosophers and wasn't familiar with some of them. And oddly, Claude helped put words to ideas I was processing. It's a mixed bag.
I appreciate that you find this an amusing pass-time.... but, spending time conversing with a machine does not appeal to me, however bright it may be? 🙃, but that aside, I do think we need to be very clear about generative AI - as in image making based on AI prompts - it is not photography, and we should not refer to it as such. There is no camera involved, no light involved, it is a purely appropriation of 1s and 0s from known sources, which are then anonymised, it is not a photograph, it never will be, and we need to stop calling it a photograph, or photography. It may be digital creation, or digiart... but, it has nothing to do with photography.
In the late 90’s I worked as an assistant for one of the better Hotel and Interior photographers on the planet. There was not many photographs that were captured that did not take 18-30 hours of lighting, styling, and composing before a couple dozen sheets of film were exposed. A common theme we made regarding the work was
“This is not reality”-
To answer the first question, yes, the camera lies all the time.
I worked on a few like that in the early90s too as a 3rd, i was having trouble staying awake it was sooooo boring. Made me question if i wanted to become a commercial photographer and i leaned into retouching pretty hard before i realized I DID wanna become a pro photographer 😂
But the pictures were, at least until dodging and burning and the like took place, proper documents of what was photographed. Why would photographs of an elaborately lit, styled set not be as real as any other photographs.
The styling, lighting, and props used to create the Images of the environment did not resemble reality. Imagine spending $30,000 in 1990's money to rent props to put in a penthouse suite, soley for the photograph, in order for it to look more elegant. Then using over a 100 lights to light literaly every inch of the environment "individually". Add working with one of the best in the world at soley doing that, along with a world class ego. Then after starting at noon the previous day, wait for a 6 minute window at sunrise to get the perfect exposure, All without so much as a wink of sleep. There was no dodging and burning involved. The lighting was perfect, the exposure was perfect. Transparency film prior to photoshop did not allow for imperfection.
But the set, however contrived, was as much a piece of reality as anything other segment of the world and the all of the effort, if anything, may have made it a more realistic, in at least one sense, depiction than, say, a blurry snapshot. Context of use might have made it a misleading depiction - if it was a staged penthouse that was up for sale it might fool a naive prospective buyer - but it still was a realistic depiction of what was photographed. Two senses of real at work here...
I've been using Claude recently as well, and what immediately stood out to me, in contrast to ChatGPT, is that Claude tends to ask questions and reflects your own words back to you in such a way that I sometimes really come to a halt—caught by clever questions, without Claude putting words in my mouth. As you already said: it can be a good catalyst for finding inspiration and arriving at a different perspective.
What's fascinating is also that these questions somehow connect with my current situation. The question about the honesty of a camera, the photographer, the situation, its representation, and the creation of a work that has permanence—this question has not let me go for weeks, ever since I began working on my first zine.
This is a brilliant experiment in what AI can synthesisze. Just being doing some reading with people like Terry Barrett and Susan Sontag and all this is right on the money so thanks a lot for the share.
I came here to have a micro rant (in good faith) about hoping that consulting with an LLM will not become a regular thing. But I see a few have beaten me to it.
I think I was prompted to respond on reading the bit where you confess to not fact checking everything.
LLMs famously make stuff up and with the web soon to create more generated content than human made stuff, we REALLY need to fact check everything we produce.
All these LLMs are doing is regurgitating what they find and that goes for other peoples slop.
If you think it's hard to trust what you find on the web now... just wait a year or two.
For the record I am a human commenter doing human things with my biological brain. ;-)
I do think we need to use these tools with a critical eye. Are we trusting machines more than humans? I'm not sure that's true. But then, more people are spreading misinformation online every day, so it's harder to find reliable information than ever. I would never rely solely on an AI, but they are helpful for idea generation and as a starting point for various lines of inquiry.
I appreciated Gary's post and generally agree. I don't use these tools often, but I find myself going to a chatbot instead of Google when I'm searching for information about an idea or topic, usually something I'm reading about. In part, it's to see what the AI says and how it responds, and also to unpack the concepts I'm studying. I am aware that there are hallucinations, etc., but I have frequently found the replies help expand my understanding of something I'm studying. Of course, I'm a nerd who regularly reads Wikipedia, just because I want to know more about something. I really appreciate the links, Christian, and the discussion. Thanks.
Thanks, Christian. I am familiar with Gary's name, but haven't read that article. I'll do so today. Here's one for you, Gideon Lewis Krause's latest for the New Yorker: What Is Claude? Anthropic Doesn’t Know, Either | The New Yorker https://share.google/hJccf4MQ57kVVpRdf
Hey, buddy, I appreciate your critique, and I expected it, which is why I did my best to be transparent about this post. It is not how I generally plan to use this space, but I found the exercise illuminating in some ways and thought it might be provocative to publish these findings. Mainly, I wanted to share what I was learning with folks like you! The AI stuff is a mixed bag, in my opinion. It’s not all gloom and doom, but there is much to dislike about it. I hope you’re well!
Cheers. All is well here. I can see this post getting ranked highly due to all these comments but don't pander to another algorithm and start posting AI images to get a rise out of us ;-)
Andy, Thanks for your post. You may be interested in my recent book - The Camera at War: 170 Years of Weaponizing Photography (https://www.barnesandnoble.com/w/the-camera-at-war-hilary-roberts/1147840266). The book is an overview of the use of photomanipulation techniques in conflict photography from the Crimea to the present day. Best wishes Hilary
Thanks, Hilary; I'll take a look. I'd love to hear more about what you're working on these days. Please email anytime at flakphoto@gmail.com. Take care!
When Michel de Montaigne needed inspiration for his Essays, he looked up to the rafters of his tower, where he had painted quotations from his favorite Latin and Greek poets, philosophers, and historians. Scholars have always surrounded themselves with books from which they drew insight and stimulus. Need no longer for this ancient practice, now that photo theory for dummies is available at the click.
Claude did an absolutely first class job of summarising the issue as economically and fair-mindedly as it is possible to do. That short piece is far more engaging than much of the flannel and emotion that has come out in the comments. Claude was just about spot on. Nice experiment, Andy. I am sure you will continue experimenting. Claude is better than many because it seems to advertise its own limitations more openly.
Claude has summarised the arguments with great clarity and economy. I would have like to see Claude incorporate the arguments of Kaja Silverman who seeks to reconceptualise photography beyond the index/construction dichotomy. In her "Miracle of Analogy" she argues that [from the blurb on Amazon.com] "photography originates in what is seen, rather than in the human eye or the camera lens, and that it is the world's primary way of revealing itself to us. Neither an index, representation, nor copy, as conventional studies would have it, the photographic image is an analogy." And she goes on to elaborate how photography, at every stage, is "developmental"
A great piece here, a conversation that I don't think will ever end in photography! I always wonder why some people perceive the camera as a truth-teller or a pathological liar in search of truth. I was taught by artists of the Sontag school of thought, hence my perspective on the camera always lying, no matter how the photographer uses it. The device can only tell the truth that it sees, and isn't that a great analogy for life!
--
I see you've already addressed a lot of people's concerns and I applaud your openness! As others have already said for themselves, I'm a bit saddened to see the use of AI even just for research — I understand the idea for what you’re trying to accomplish for the sake of this piece, to get its algorithmically synthesized answer to this question. I don't want to come off as accusatory or angry. I even get sad when friends of mine are *forced* to use AI for work, and I don't blame them, either! You explained your purpose for using AI upfront and transparently, which is essential these days -- and I thank you for that!
AI is a tool, yes, but I've been abundantly public about my perspective on AI being an evil technology, so my bias is massively at play here. I see this technology as actively destroying society, and yes, your application here is pretty harmless at large. But I don't think I'll ever *not* see AI as a machine that's designed to displace human thought.
My own feelings be damned, I can't control what you do!
Cheers, Jeff. I appreciate your perspective and think it's wise to be wary of this technology. I'm always as curious about digital media as I am about photography, and I probably buried that lede a bit here.
Part of my interest is seeing how Claude responds to the question and how that compares with how many human photographers responded to the same question on social media in previous days.
I very much like photography theory, so this was right up my alley, and it gave me some new avenues for inquiry. I am mixed about AI and have previously written about AI imagery here. You might like this piece from a few years ago.
The summary Claude produced is nice but in my opinion not complete. It misses the question of what “lie” and “truth” is. If you replace, in Claude’s narrative, the word “photo” by documentary, movie, painting, news article, novel, research, it would still have the same message. In fact any human observation is done from a point-of-view, is framed, is adjusted to or biased by local conditions, is filtered by opinions, is decided by picking a moment and is a product of selection. So the question is whether any human observation can be entirely true, in a sense that it produces a God’s eye view of reality. Photography has the same value and shortcomings as any observation. And there’s no sense in making it inferior to text, thoughts, art or other ways of human expression.
I’m not sure what all the fuss is about. Essentially, you asked AI to put together a brief compilation of what people have said about the authenticity of photographs, and it gave you a useful summary. I read your piece and thought, “This is the kind of thing that makes me think, ‘Hmmm…’” Thanks. (By the way, I put the question, “How many angels can fit on the head of a pin?” to ChatGPT, and the answer was, “As many as you like.” Ha! If you didn’t know, now you know.)
dont mean to be harsh, but i hope this AI generated stuff can go into a different feed and not the one i subscribe to
Hey Nathan! It's not harsh. I don't intend to make this a habit and I appreciate the note. I hope you're well!
I'm not a fan of AI either; it's simply generating concepts and arguments made by others. That said, the points raised are valid and worth discussing.
I hear you, and thanks for the note. I worried this might put some people off — and for good reason. I was inspired by the ideas and wanted to share them, so I figured I'd take a risk and put this out there. Thanks, hw.
I am not sure AI is so different from us. Most people default to generic answers unless they are encouraged to go further. If you ask a shallow question, you receive a shallow response, whether from AI or a person. However, when you challenge it, reformulate, and question the question itself, the discussion deepens.
I see this frequently in teaching. I have 56 students, and most of them initially repeat generic ideas. What they have read, what they have heard, what seems logical to say to avoid sounding foolish, or what someone in authority has already stated. Real thinking begins when you ask, “Is this actually my view? What can I add from my own experience or observations?”
In that sense, AI feels more like a mirror. The depth depends on how willing we are to think, and how much we are prepared to push it, or ourselves, further.
Hey Nathan, there are plenty of reasons to take issue with AI. I certainly do for many reasons. It's also probably in our best interest to not be naive about the capabilities of these technologies. Claude has managed to answer a photo 101 paradox using resources most photographers will never read. This post and the answer Claude provided is a philosophical question in itself. Considering substack was started as a platform for writing, it's probably in our best interest to know that AI can answer a foundational photography issue better than most photographers. Personally, I would like to ignore AI all together. I think it's likely that the broligarchy is going to try to use it to take our money and attention away. It's going to be making perfect realistic pictures too. We are crossing a threshold and without understanding what it is capable of we can't begin to know where or how to move forward. I wish this had not happened during my lifetime, but I cannot deny that Claude’s response here seems to address this very issue in the last sentence; “The question “Does the camera lie?” might ultimately be less interesting than the question it opens up: What do we want from truth, and why did we ever think a machine could give it to us?”
Hey, Randall, these are all excellent points. And I agree with that last statement. I find myself asking Claude questions from time to time, mainly to inspire reflection and to help me work through my thoughts. I had a feeling that doing this in public might catalyze some community conversations. I hope it does.
I believe tech companies are coming for everything we do, such that in the future there will hardly be room for human interaction that isn't filtered through the distorting lens of AI. And so I am currently operating with the theory that every inch I give to the LLMs is space taken away, exponentially, from humans. And I want to engage with people thinking and saying and doing human things. So — successful provocation aside — it's a no from me, dawg.
You know I'm a fan, Bill, so I appreciate your feedback. Fear not, this won't be a recurring feature! Still, the ideas are compelling, and I know you appreciate them. I learned something new from these philosophers and wasn't familiar with some of them. And oddly, Claude helped put words to ideas I was processing. It's a mixed bag.
I appreciate that you find this an amusing pass-time.... but, spending time conversing with a machine does not appeal to me, however bright it may be? 🙃, but that aside, I do think we need to be very clear about generative AI - as in image making based on AI prompts - it is not photography, and we should not refer to it as such. There is no camera involved, no light involved, it is a purely appropriation of 1s and 0s from known sources, which are then anonymised, it is not a photograph, it never will be, and we need to stop calling it a photograph, or photography. It may be digital creation, or digiart... but, it has nothing to do with photography.
I completely agree, Soren.
In the late 90’s I worked as an assistant for one of the better Hotel and Interior photographers on the planet. There was not many photographs that were captured that did not take 18-30 hours of lighting, styling, and composing before a couple dozen sheets of film were exposed. A common theme we made regarding the work was
“This is not reality”-
To answer the first question, yes, the camera lies all the time.
I worked on a few like that in the early90s too as a 3rd, i was having trouble staying awake it was sooooo boring. Made me question if i wanted to become a commercial photographer and i leaned into retouching pretty hard before i realized I DID wanna become a pro photographer 😂
But the pictures were, at least until dodging and burning and the like took place, proper documents of what was photographed. Why would photographs of an elaborately lit, styled set not be as real as any other photographs.
The styling, lighting, and props used to create the Images of the environment did not resemble reality. Imagine spending $30,000 in 1990's money to rent props to put in a penthouse suite, soley for the photograph, in order for it to look more elegant. Then using over a 100 lights to light literaly every inch of the environment "individually". Add working with one of the best in the world at soley doing that, along with a world class ego. Then after starting at noon the previous day, wait for a 6 minute window at sunrise to get the perfect exposure, All without so much as a wink of sleep. There was no dodging and burning involved. The lighting was perfect, the exposure was perfect. Transparency film prior to photoshop did not allow for imperfection.
But the set, however contrived, was as much a piece of reality as anything other segment of the world and the all of the effort, if anything, may have made it a more realistic, in at least one sense, depiction than, say, a blurry snapshot. Context of use might have made it a misleading depiction - if it was a staged penthouse that was up for sale it might fool a naive prospective buyer - but it still was a realistic depiction of what was photographed. Two senses of real at work here...
I've been using Claude recently as well, and what immediately stood out to me, in contrast to ChatGPT, is that Claude tends to ask questions and reflects your own words back to you in such a way that I sometimes really come to a halt—caught by clever questions, without Claude putting words in my mouth. As you already said: it can be a good catalyst for finding inspiration and arriving at a different perspective.
What's fascinating is also that these questions somehow connect with my current situation. The question about the honesty of a camera, the photographer, the situation, its representation, and the creation of a work that has permanence—this question has not let me go for weeks, ever since I began working on my first zine.
This is a brilliant experiment in what AI can synthesisze. Just being doing some reading with people like Terry Barrett and Susan Sontag and all this is right on the money so thanks a lot for the share.
I came here to have a micro rant (in good faith) about hoping that consulting with an LLM will not become a regular thing. But I see a few have beaten me to it.
I think I was prompted to respond on reading the bit where you confess to not fact checking everything.
LLMs famously make stuff up and with the web soon to create more generated content than human made stuff, we REALLY need to fact check everything we produce.
All these LLMs are doing is regurgitating what they find and that goes for other peoples slop.
If you think it's hard to trust what you find on the web now... just wait a year or two.
For the record I am a human commenter doing human things with my biological brain. ;-)
But people also famously make things up... And we also depend on hearsay for many of our beliefs.
Yes of course. I've read fiction. And marketing :-) It's just that at the moment we appear to be trusting machines more than humans.
I do think we need to use these tools with a critical eye. Are we trusting machines more than humans? I'm not sure that's true. But then, more people are spreading misinformation online every day, so it's harder to find reliable information than ever. I would never rely solely on an AI, but they are helpful for idea generation and as a starting point for various lines of inquiry.
Have you read Gary Marcus? This is a good start... https://garymarcus.substack.com/p/stop-treating-ai-models-like-people This Forbes piece on how in the workplace Gen Z trust AI over humans is also worth a read. https://www.forbes.com/sites/bryanrobinson/2025/02/19/gen-z-trust-ai-over-humans-in-their-careers-new-study-shows/ Interesting times for sure.
I appreciated Gary's post and generally agree. I don't use these tools often, but I find myself going to a chatbot instead of Google when I'm searching for information about an idea or topic, usually something I'm reading about. In part, it's to see what the AI says and how it responds, and also to unpack the concepts I'm studying. I am aware that there are hallucinations, etc., but I have frequently found the replies help expand my understanding of something I'm studying. Of course, I'm a nerd who regularly reads Wikipedia, just because I want to know more about something. I really appreciate the links, Christian, and the discussion. Thanks.
Thanks, Christian. I am familiar with Gary's name, but haven't read that article. I'll do so today. Here's one for you, Gideon Lewis Krause's latest for the New Yorker: What Is Claude? Anthropic Doesn’t Know, Either | The New Yorker https://share.google/hJccf4MQ57kVVpRdf
Hey, buddy, I appreciate your critique, and I expected it, which is why I did my best to be transparent about this post. It is not how I generally plan to use this space, but I found the exercise illuminating in some ways and thought it might be provocative to publish these findings. Mainly, I wanted to share what I was learning with folks like you! The AI stuff is a mixed bag, in my opinion. It’s not all gloom and doom, but there is much to dislike about it. I hope you’re well!
Cheers. All is well here. I can see this post getting ranked highly due to all these comments but don't pander to another algorithm and start posting AI images to get a rise out of us ;-)
I can really recommend this podcast should you want a picture of the current state of AI at the moment... https://www.bbc.co.uk/sounds/brand/m001wjf8
Ha. I know, AI ruffles photography feathers. Thanks for the link. I'll check it out.
Andy, Thanks for your post. You may be interested in my recent book - The Camera at War: 170 Years of Weaponizing Photography (https://www.barnesandnoble.com/w/the-camera-at-war-hilary-roberts/1147840266). The book is an overview of the use of photomanipulation techniques in conflict photography from the Crimea to the present day. Best wishes Hilary
Thanks, Hilary; I'll take a look. I'd love to hear more about what you're working on these days. Please email anytime at flakphoto@gmail.com. Take care!
Yeah, A\ is doing the best job. As a Technologist, I am trying to work for them :)
When Michel de Montaigne needed inspiration for his Essays, he looked up to the rafters of his tower, where he had painted quotations from his favorite Latin and Greek poets, philosophers, and historians. Scholars have always surrounded themselves with books from which they drew insight and stimulus. Need no longer for this ancient practice, now that photo theory for dummies is available at the click.
Now the question is, does AI lie?
Claude did an absolutely first class job of summarising the issue as economically and fair-mindedly as it is possible to do. That short piece is far more engaging than much of the flannel and emotion that has come out in the comments. Claude was just about spot on. Nice experiment, Andy. I am sure you will continue experimenting. Claude is better than many because it seems to advertise its own limitations more openly.
Claude has summarised the arguments with great clarity and economy. I would have like to see Claude incorporate the arguments of Kaja Silverman who seeks to reconceptualise photography beyond the index/construction dichotomy. In her "Miracle of Analogy" she argues that [from the blurb on Amazon.com] "photography originates in what is seen, rather than in the human eye or the camera lens, and that it is the world's primary way of revealing itself to us. Neither an index, representation, nor copy, as conventional studies would have it, the photographic image is an analogy." And she goes on to elaborate how photography, at every stage, is "developmental"
A great piece here, a conversation that I don't think will ever end in photography! I always wonder why some people perceive the camera as a truth-teller or a pathological liar in search of truth. I was taught by artists of the Sontag school of thought, hence my perspective on the camera always lying, no matter how the photographer uses it. The device can only tell the truth that it sees, and isn't that a great analogy for life!
--
I see you've already addressed a lot of people's concerns and I applaud your openness! As others have already said for themselves, I'm a bit saddened to see the use of AI even just for research — I understand the idea for what you’re trying to accomplish for the sake of this piece, to get its algorithmically synthesized answer to this question. I don't want to come off as accusatory or angry. I even get sad when friends of mine are *forced* to use AI for work, and I don't blame them, either! You explained your purpose for using AI upfront and transparently, which is essential these days -- and I thank you for that!
AI is a tool, yes, but I've been abundantly public about my perspective on AI being an evil technology, so my bias is massively at play here. I see this technology as actively destroying society, and yes, your application here is pretty harmless at large. But I don't think I'll ever *not* see AI as a machine that's designed to displace human thought.
My own feelings be damned, I can't control what you do!
Looking forward to the next piece, Andy!
Cheers, Jeff. I appreciate your perspective and think it's wise to be wary of this technology. I'm always as curious about digital media as I am about photography, and I probably buried that lede a bit here.
Part of my interest is seeing how Claude responds to the question and how that compares with how many human photographers responded to the same question on social media in previous days.
For example: https://www.instagram.com/p/DUnihWSDqtQ/
I very much like photography theory, so this was right up my alley, and it gave me some new avenues for inquiry. I am mixed about AI and have previously written about AI imagery here. You might like this piece from a few years ago.
https://www.flakphoto.news/p/its-not-photography
Thanks for the note! It's great to hear from you.
The summary Claude produced is nice but in my opinion not complete. It misses the question of what “lie” and “truth” is. If you replace, in Claude’s narrative, the word “photo” by documentary, movie, painting, news article, novel, research, it would still have the same message. In fact any human observation is done from a point-of-view, is framed, is adjusted to or biased by local conditions, is filtered by opinions, is decided by picking a moment and is a product of selection. So the question is whether any human observation can be entirely true, in a sense that it produces a God’s eye view of reality. Photography has the same value and shortcomings as any observation. And there’s no sense in making it inferior to text, thoughts, art or other ways of human expression.
I’m not sure what all the fuss is about. Essentially, you asked AI to put together a brief compilation of what people have said about the authenticity of photographs, and it gave you a useful summary. I read your piece and thought, “This is the kind of thing that makes me think, ‘Hmmm…’” Thanks. (By the way, I put the question, “How many angels can fit on the head of a pin?” to ChatGPT, and the answer was, “As many as you like.” Ha! If you didn’t know, now you know.)