AI: the Worse Enemy of Photography

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

Perhaps I am redundant because some time ago I posted a couple of threads here about my dislike – or rather – my rejection, of AI applied to photography.
I remember that the majority of comments I received in response were critical and maintained that the advancement of technology should be accepted and applauded.
In those previous cases I made reference to several examples of how prizes have been awarded to photographs that turned out to be altered or directly created with this technology, now, with the false and altered photos of British royalty, the issue has taken on global relevance, which is why Photographers must take note.
It would be interesting to read what the defenders of AI think now, which for me is the worst enemy of photography, that is, one of the biggest traps of our times. We will never know what is true or a lie.
Kind regards to all.
 
Which is worse: a computer algorithm alters a photo to make it more fantastic, or a human doing so?

Most of us play worth Lightroom sliders. Many of us use Photoshop to make content changes. Some of us composite entire scenes, to the point they no longer resemble the original scene.

Human editing is generally accepted outside reportage. Wipe out power lines behind a BIF phot and people will generally congratulate your photo. But what if I could press a button and have my PC do the job in 1/10th the time it took me in Photoshop?

Is automation the problem?
 
Walter has a valid point here. The long gone days of “dodging” & “burning-in” in a darkroom have been replaced with technology and a computer. How far is too far? Recent cell phone ads are promoting a photo AI that quickly removes anything from an image. That includes improving complexions and features on peoples faces. I think we’re witnessing a slow moving train wreck to the talents and skill sets of photographers…….😕
 
  • Like
Reactions: seh
Perhaps I am redundant because some time ago I posted a couple of threads here about my dislike – or rather – my rejection, of AI applied to photography.
I remember that the majority of comments I received in response were critical and maintained that the advancement of technology should be accepted and applauded.
In those previous cases I made reference to several examples of how prizes have been awarded to photographs that turned out to be altered or directly created with this technology, now, with the false and altered photos of British royalty, the issue has taken on global relevance, which is why Photographers must take note.
It would be interesting to read what the defenders of AI think now, which for me is the worst enemy of photography, that is, one of the biggest traps of our times. We will never know what is true or a lie.
Kind regards to all.
Remember the old Kremlin photos that removed individuals who were now considered "enemies of the people." Now it's just easier, and yes, more insidious and dangerous. I agree. But AI used for noise removal and sharpening is terrific in my estimation. Yeah, people will take any technology and use it for bad ends. Not sure if there is a solution. No I am sure, there is not.
 
I think it will bolster the value of those artists that can actually capture the various things that AI can easily produce. Sure AI can generate a bald eagle in front of the total eclipse, but the artist that gets a real one and can prove it will be well regarded.
 
IMO, there are 2 types of photography uses of AI now. One, like sharpening, noise removal, keeps the photo as taken, just refines it. Second, is creating new images or parts of images into an existing photo or creating a whole new photo, adding skies, etc.. Then there is a blend of where the two mix. 2 different purposes, both with a role. Like painting, photos can now be created and manipulated to express the creators vision, which may or may not reflect what was seen through the view finder. The importance is that two be indicated for what they are, i.e., an AI generated photo should be indicated as such, not trying to be passed off as a 'real life' photo. A lot of difference between the person sitting at a computer creating the eagle in front of an eclipse, vs the person actually working to set up and get the photo in real time (to use bleirer's example). Although someone viewing the photos may not be able to distinguish the difference.
 
Perhaps I am redundant because some time ago I posted a couple of threads here about my dislike – or rather – my rejection, of AI applied to photography.
I remember that the majority of comments I received in response were critical and maintained that the advancement of technology should be accepted and applauded.
In those previous cases I made reference to several examples of how prizes have been awarded to photographs that turned out to be altered or directly created with this technology, now, with the false and altered photos of British royalty, the issue has taken on global relevance, which is why Photographers must take note.
It would be interesting to read what the defenders of AI think now, which for me is the worst enemy of photography, that is, one of the biggest traps of our times. We will never know what is true or a lie.
Kind regards to all.
Like many controversial topics, I don't think there is a simple "Good/Bad" or "Yes/No" answer to this. I would be against AI being used to generate images from scratch and then being passed off as photographs taken in the "wild"...I am not against AI being used, as mentioned in a previous post, to help me intelligently apply sharpening or to remove noise. I would be against AI being used with the intent of deception...I am not against the use of AI as it applies to the algorithms being used to assist in the accuracy of AF (bird detection, people detection, eye detection, etc.). AI and its applications remind me of Pandora's Box...the box has been opened, there's not shoving it back in...its up to us to use it intelligently, ethically and morally.
 
I am very much against AI. I can see minor use cases, yes, and I dont see an issue with removing minor things in photos. It's when you use AI to completely replace things where I have the problem.

You used to have to get up early, drive an hour, hike for an hour, set up, and wait for the sun to get that perfect sunrise. But oops, too cloudy, try again another day.
Now? Hike up in the middle of the day, who cares, use AI to put a fake sunrise in.

The former, while I can agree can be frustrating and annoying, is part of the fun and satisfaction when you nail the shot. Now you can cheat to get it and pretend to the world you did the former when you really didn't.
 
I have a suspicion that reasonably soon we'll see cameras enact features to support the Content Authenticity Initiative (CAI) / Coalition for Content Provenance and Authenticity (C2PA) as well as editing tools like those from Adobe such that it will be accessible to be able to see in an authoritative way the entire editing history of an image starting with a verification of the camera that took the image all the way to any modifications made to it and it will be common for professional sources like new agencies, but also any professional photography to participate in the use of these tools.

That said, of course, that doesn't take away from the FUD of photos that don't use these tools, and of course, many people aren't critical of what they see (or hear or read) anyway.

And personally, I plan to adopt use of these tools as they become available to me.
 
I think that this is certainly a multifaceted issue and there are uses of AI which are more "legitimate" or "honest" than others.

Not having heard anything about the Royal family, I looked it up and think it's a helpful example. I came across stories about two family photos that have drawn a great deal of flak for having been edited, but the thing is that the sorts of edits in question are the sorts of edits portrait photographers do every day and have been doing every day since long before Photoshop existed. To me this is clearly in the realm of artistic expression. Yet many people out there don't understand this - they think their wedding photos, for instance, show what it really looked like that day - and so they demonize stuff like this.

On the other hand, the potential for abuse requires no explanation and yes AI technology does make it not only easier but opens up almost infinitely many new possibilities for exactly what sorts of abuse can be practiced. It also does seem to me to be very much a freight train coming straight at many categories of professional photographer. People who shoot sports and events are probably safe, at least for now, but before too long what will be the point of spending a few hundred dollars on family portraits when the click of a button can generate them in seconds from your iPhone gallery? Why would any company spend a cent on a skillfully captured wildlife shot when they can instantly generate one that perfectly suits their purposes for the tiniest fraction of the cost?

This is, by the way, really quite different from the perceived challenge that painting saw with the advent of photography. The overall look and style of paintings vs. photographs have always been quite markedly different - until today, that is.

In any case, I could write quite extensively on this closing point, but to try to be a bit concise I will say that I think a time is coming and really quite soon when we as human beings are going to be faced with a choice [- and not only as pertains to photography but as pertains to many, many things - about just how much we value human dignity as a good and as an end unto itself. We'll soon reach a point when the only value to human work over what has been generated by a computer will be the very fact that it was done by a human being. If we decide that all we care about is results, I think our society will devolve into something very much resembling the society of the Axiom spacecraft in the film Wall-E. If we decide that human achievement is worthwhile as an end unto itself - to me the far better choice - we'll have to have the strength to knowingly choose "worse" options just to value the good of humanity.

I don't know how it will shake out in the end, but I do think that at least for a little while this will probably look a little something like the dynamic we have now with large corporations and small businesses, where many people will choose the AI stuff for convenience and cost-effectiveness while a significantly smaller group of people will choose "lower quality" and more expensive human artistic work but do so on principle. In other words, industries like photography will survive but they will be much, much smaller with far fewer paying clients to try to win over.
 
I'm interested in a lot of the good aspects of AI. For example, right now we have exceptionally good subject identification for birds using Merlin Photo AI - and Merlin Sound AI for identification of sounds and calls. Swarovski has already integrated the Photo ID with electronic binoculars so you can press a button on the binoculars and have bird identification. I suspect we are very close to an app where you can automate Photo ID using your computer - and possible with your camera. It's a short step to link that with populating the subject name, scientific name, and other information in the keywords.

The use of AI based tools for subject recognition has a lot of power. It's already being used in our cameras to improve AF. It's also being used by software to improve sharpening and noise reduction. At times it also can enhance photos with texture such as fur and feathers - and I've seen it miss badly here. But it will improve.

I've heard about AI support in sports similar to the birding applications above. One application is the ability to track subjects by jersey number - something being used already. Another is potentially the ability to keyword images with the team name, player name, and position using AI plus a list of players and numbers for each team.

AI is also being used to speed up image triage after an event. Zenfolio has an AI based product that does a pretty good job of going through portraits and making selections based on facial expression and eyes. The product can also be used for landscapes where it does a satisfactory job. For wildlife, it's not really there yet, but it will improve. This tool was more than 90% accurate in my testing for people and events.

When it comes to new content in an image. I think the necessary inclusion of metadata indicating what tools were used will solve some problems. New content in an image is potentially problematic - but could be useful and appropriate for artistic and commercial uses. These edits are possible without AI - just a lot easier with AI tools.
 
I don't distinguish between human or AI.

Photography for several decades has not been about the gear, the shot, or the photographer. Editing is the most important part in producing an appealing photo. And as soon as anyone takes that raw image and puts it into a computer, it's no longer "original" or "real".

Like everything in life, you will have people who try to "cheat" the system (use AI in competitions). It's more a question of ethics than AI.

Nowadays photography should mainly be about fun. Artists would starve even more today than they would in prior decades, if they resorted to only selling prints. That's why you see the "top" photographers making a living off workshops, 1:1 sessions, editing sessions, subscription setups, youtube content, etc. and personally - I prefer it this way. Get rid of anyone who wants to make art just for profit, and leave only the people who care about it.

When I look at a good image I don't wonder "what percent of this is real life, human, versus editing and AI" I look at and I think "does this image appeal to me? do I like the subject? the story it tells? the technique?"

just my .02
 
Last edited:
I'm interested in a lot of the good aspects of AI. For example, right now we have exceptionally good subject identification for birds using Merlin Photo AI - and Merlin Sound AI for identification of sounds and calls. Swarovski has already integrated the Photo ID with electronic binoculars so you can press a button on the binoculars and have bird identification. I suspect we are very close to an app where you can automate Photo ID using your computer - and possible with your camera. It's a short step to link that with populating the subject name, scientific name, and other information in the keywords.

The use of AI based tools for subject recognition has a lot of power. It's already being used in our cameras to improve AF. It's also being used by software to improve sharpening and noise reduction. At times it also can enhance photos with texture such as fur and feathers - and I've seen it miss badly here. But it will improve.

I've heard about AI support in sports similar to the birding applications above. One application is the ability to track subjects by jersey number - something being used already. Another is potentially the ability to keyword images with the team name, player name, and position using AI plus a list of players and numbers for each team.

AI is also being used to speed up image triage after an event. Zenfolio has an AI based product that does a pretty good job of going through portraits and making selections based on facial expression and eyes. The product can also be used for landscapes where it does a satisfactory job. For wildlife, it's not really there yet, but it will improve. This tool was more than 90% accurate in my testing for people and events.

When it comes to new content in an image. I think the necessary inclusion of metadata indicating what tools were used will solve some problems. New content in an image is potentially problematic - but could be useful and appropriate for artistic and commercial uses. These edits are possible without AI - just a lot easier with AI tools.

Unfortunately I don't really think metadata is much of a solution to anything. I cannot imagine that it will be too long before methods exist to defeat or spoof it, for instance. Even if that weren't to happen, I don't see how it will really matter for many of the uses that this kind of system would be most important for. Sure, the AP could verify whether a photo from an NFL game or some newsmaking event is authentic, but ultimately if some unauthenticated photo of politician X doing thing Y with accomplice Z pops up, it's going to spread around the internet and have just as much impact on people as it would if it were an authenticated photo. This is ultimately what already happens with social media, after all: people see an article or link or meme, they share it, their friends share it, and soon everyone believes it regardless of whether it's true or not. I don't see how authentication fixes much of anything.

Heck, I can see ways it might make things worse. I can image there will be wedding photographers who will edit their photos the way they always have suddenly getting angry clients upset that their photos report they've been altered, for instance.
 
I'm interested in a lot of the good aspects of AI. For example, right now we have exceptionally good subject identification for birds using Merlin Photo AI - and Merlin Sound AI for identification of sounds and calls. Swarovski has already integrated the Photo ID with electronic binoculars so you can press a button on the binoculars and have bird identification. I suspect we are very close to an app where you can automate Photo ID using your computer - and possible with your camera. It's a short step to link that with populating the subject name, scientific name, and other information in the keywords.

The use of AI based tools for subject recognition has a lot of power. It's already being used in our cameras to improve AF. It's also being used by software to improve sharpening and noise reduction. At times it also can enhance photos with texture such as fur and feathers - and I've seen it miss badly here. But it will improve.

I've heard about AI support in sports similar to the birding applications above. One application is the ability to track subjects by jersey number - something being used already. Another is potentially the ability to keyword images with the team name, player name, and position using AI plus a list of players and numbers for each team.

AI is also being used to speed up image triage after an event. Zenfolio has an AI based product that does a pretty good job of going through portraits and making selections based on facial expression and eyes. The product can also be used for landscapes where it does a satisfactory job. For wildlife, it's not really there yet, but it will improve. This tool was more than 90% accurate in my testing for people and events.

When it comes to new content in an image. I think the necessary inclusion of metadata indicating what tools were used will solve some problems. New content in an image is potentially problematic - but could be useful and appropriate for artistic and commercial uses. These edits are possible without AI - just a lot easier with AI tools.
All the items you mention have nothing to do with photography as such, what you refer to is with the search, identification, information, selection and archiving of subjects or things – in this case, birds – and not necessarily applied to photography, since these could be used for scientific purposes or simply for entertainment, for example. I approach the subject strictly from photography, which specifically consists of capturing what the photographer sees when shooting. However, perfectly legitimate tools such as improving color tone, adjusting focus, removing noise or brightening a scene using PS is nothing more than the current digital mode of editing, and from my point of view, it is very far from being AI, whose function and application is to intervene and falsify something that was never on camera.
 
I cannot imagine that it will be too long before methods exist to defeat or spoof it, for instance.
i don't think this is a huge risk, at least not for a while. we are actually quite accomplished with cryptography at this point in our history and while eventually there will be problems with older authentication technology, it's likely going to lag and be most of a problem with old content which will by that time, be "history" so to speak.

Sure, the AP could verify whether a photo from an NFL game or some newsmaking event is authentic, but ultimately if some unauthenticated photo of politician X doing thing Y with accomplice Z pops up, it's going to spread around the internet and have just as much impact on people as it would if it were an authenticated photo.
this is the big issue of course.

i think it is likely that it will become a requirement for all the news outlets and the norm for professionals in general in short order once technology becomes reasonably available (and i think we are close to that). thus, the "norm" for anyone wanting to be taken seriously.

but of course what you say is the problem. we the public tend not to be critical when evaluating content and unauthenticated will be introduced and be treated as if it were.

in addition, we have the problem that you can still stage scenes. obviously that will be increasingly tricky if the authentication encapsulates gps and timestamps (which i'm sure it will), but then people will still work around it, for example, blocking their gps and then still claim it's authenticated.

if we want truth, we're going to have to fight for it
 
My concern is that every photo will have to be perfect in order to be acceptable....

...by me...

...as I compare my work to the constantly evolving standards of "goodness" and, in the process, strip away the joy I derive from photography.
 
  • Like
Reactions: seh
My concern is that every photo will have to be perfect in order to be acceptable....

...by me...

...as I compare my work to the constantly evolving standards of "goodness" and, in the process, strip away the joy I derive from photography.
this is already and has been a problem.

we start out and we compare ourselves to those who are crazy talented, and compare our every photo to their BEST photos.

then add on top of that any editing (let's anchor that to Ansel Adams, just so we don't think that's a "new" thing).

our monkey brain says "ooh, that's great, i want to do THAT". then we try to do that, perhaps not really internalizing that what we're comparing to is both the apex of accomplishment, but also perhaps... not real at all
 
i don't think this is a huge risk, at least not for a while. we are actually quite accomplished with cryptography at this point in our history and while eventually there will be problems with older authentication technology, it's likely going to lag and be most of a problem with old content which will by that time, be "history" so to speak.

I think what it will come down to is that there is probably going to be a need for some kind of compromise between a truly difficult to spoof or crack system and one that can function with the sort of agility across multiple platforms and a lot of different sorts of standalone devices. You could make an extremely difficult to beat system, but I'm not so sure you can do so in a way that works within the sorts of practical realities necessary for working photographers.

The other issue is this: if this is going to work, it's going to mean there will have to be a standard for how these authentication keys or whatever work - but if there's a standard then that means that anyone with the sufficient coding skills could just create software to create keys that follow that standard.
this is the big issue of course.

i think it is likely that it will become a requirement for all the news outlets and the norm for professionals in general in short order once technology becomes reasonably available (and i think we are close to that). thus, the "norm" for anyone wanting to be taken seriously.

but of course what you say is the problem. we the public tend not to be critical when evaluating content and unauthenticated will be introduced and be treated as if it were.

in addition, we have the problem that you can still stage scenes. obviously that will be increasingly tricky if the authentication encapsulates gps and timestamps (which i'm sure it will), but then people will still work around it, for example, blocking their gps and then still claim it's authenticated.

if we want truth, we're going to have to fight for it
Another issue here is with social media sites and the way they handle images. Facebook, Instagram, presumably Twitter (or X, or whatever it is), etc. don't just post the original files people upload, but take them and process them into altogether different photos, stripping the metadata in the process. This is probably true of most media sites as well. They could in theory change the way things work so that metadata is preserved, but this then gets into all sorts of questions about whether any functional authentication system would be able to allow "transfers" or the authentication.
 
Back
Top