luxBorealis Blog

AssistiveAI, GenAI, Internal, External—Where’s the Authenticity?

NOTE: This article was published simultaneously on the Luminous-Landscape.com

AI is not photography! Or is it?

Let’s get once thing out of the way right off the top—the argument regarding AI should not be about ‘honesty’. Photography lost that battle decades ago. It has never truly been ‘honest’, though many still perceive it as so. Take a picture of your family and you get a reasonable facsimile on screen or in a print—complete with goofy looks, hair sticking up and that spare tire you’re carrying around.

But photos have been faked forever, and it’s not just Bigfoot and the Loch Ness Monster. From Abraham Lincoln’s portrait, to National Geographic’s Pyramids cover, even World Press winning photographs, hoaxes, misrepresentations, and alterations have existed for almost as long as the medium itself. Have a look at The Hoax Museum Photo Archive and you’ll get the idea.

So, AI doesn’t really change things, or does it?

Maybe we need to reframe how we think of AI. To my mind,

AI is to authentic photography as ultra-processed foods are to real food.

Ultra-processing reduces costs and make life more convenient, but a regular diet of it makes us lazy and less healthy.

AI is much the same. AI reduces costs and makes life more convenient. And, yes, it will also make photographers lazy and possibly less healthy.

Any foodie will understand the difference between real food and the ultra-processed stuff that has crept into our grocery carts. Think about all those artificial ingredients and the unnecessary fats, sugars and starches that are added to food to make you crave more of it. To the average consumer, UPF is convenient and it tastes great. The fact that it leads to health problems, such as the obesity and diabetes, doesn’t seem to matter to most people. AI in photography is much the same.

Assistive AI

To be fair, not all AI is a problem in photography. Various types of Assistive AI are commonplace. Most phone and camera manufacturers have incorporated Assistive AI and machine learning in some way, most often for AF-assist, but also for in-camera processing and up-scaling. Algorithms, such as Adobe’s Sensei, help you find photos in Lightroom without keywording. Other apps analyze photos and suggest keywords. Assistive AI also streamlines many tasks like creating background and subject masks. And, for high volume photographers of sports teams, events, grad photos, etc. it’s a real time-saver. But Assistive AI is not what people are concerned about. There is another aspect to AI that is raising concerns.

Generative AI

The difference with GenAI is that it creates new pixels in an image. But even GenAI comes in two flavours: internal and external. Internal GenAI analyzes the pixels within the image, then uses them to create pixels, kind of like cloning but more automated and on steroids. On the other hand, External GenAI uses AI algorithms to create pixels entirely new to the image, taking them from third-party photographs—photos that are not your own.

You’ll know this from Photoshop’s Generative Fill. It uses pixels from other photographers’ work to fill in backgrounds and skies, to remove objects, and to fill in gaps and edges. You may have noticed ‘Generative Credits’ as part of your subscription. Those are to ‘pay for’ External GenAI services. Topaz Labs clearly labels and promotes their PhotoAI app as one that will generate pixels to fill in missing details. They also offer the option of using Cloud Rendering, for which they will sell you credits. And Luminar Neo boasts its GenErase, GenSwap and GenExpand. 

With GenAI creeping into our processing, photographers need to be aware of exactly what their tools are doing and how GenAI is working, especially when selling work, copyrighting it, or entering photo contests and competitions. In most cases, especially for personal use, it’s no problem at all—full GenAI ahead. But for other uses, it’s important to know which GenAI is permitted and which is not.

Part of the confusion lies with a lack of clarity in labelling. As mentioned above, Internal GenAI assesses the pixels within an image to make adjustments. You may be familiar with Adobe’s new ‘Distraction Removal – Reflections’ option. It works entirely internally, by analyzing existing pixels using internal AI algorithms, so it is not introducing any new pixels to the image. Many Denoise and Sharpen adjustments work internally as well, for example DxO Pure Raw and Lightroom’s ‘Enhance’ feature. Adobe’s Neural Filters are another example, but only some of them use Internal GenAI, for example, ‘Colorize’, ‘JPEG Artefact Removal’, and ‘Skin Smoothing’. So while all of this is GenAI, it is not using third-party images, so they are all ‘safe’ to use.

But is it ‘yours’?

The confusion begins as some of Adobe’s Neural Filters use External GenAI, introducing entirely new pixels to your photograph based on the works of other photographers. These include ‘Back Drop’, ‘Deep Blur’ and ‘Make-up Transfer’, to name a few.

So, the question arises, if you drop in a background taken from another photographer’s photo** is it still ‘your’ photo? After all, the new pixels aren’t your pixels—it was all done by computer algorithms based on photos you didn’t make. Copyright is based on ‘original works’ created by a ‘human author’. Is it original? Is it even human? ‘But—’ you might say, ‘it was my idea, my concept, right?’ Unfortunately, ideas can’t be copyrighted, only the physical expression of an idea.

**BTW: This reality of ‘taken from another photographer’s photo’ opens up another hornet’s nest of controversy. GenAI has to ‘learn’ and it needs a ‘pool’ of images to work with and ‘borrow’ from. Programmers have used photos from all over the web to do that, usually without the direct permission of photographers. In the wild west of fine print in user agreements, your photos are likely part of that learning (LINK). 

If the image you produce isn’t entirely yours, nor was it generated completely by you, should you be taking the credit for it? Are you faking ownership if you put it forth as ‘your’ work? In once sense, it’s similar to ‘Made in Canada’ versus ‘Product of Canada’. ‘Made in’ only needs 51% of the content to be Canadian. ‘Product of’ requires 98%.

Recently the US Copyright Department has tried to clarify things. Interestingly, a hundred years ago, photographs themselves had to battle to be accepted for copyright. At the time, many felt that because photographs were machine-made they should not be. Now that we’re long over that hurdle, it seems the human touch remains a contributing factor. A human-made photograph is copyrightable; a machine-generated image, even if made using human commands, is not. The jury is still out on photographs that use GenAI to create parts of the image. Time will tell.

Incongruencies 

Then there is the visual impact. There has always been poor photography and there are times when GenAI doesn’t quite hit the mark. And worse, there are instances when the user thinks, “Wow! This looks amazing!”, but others can see right through the attempt at making something from nothing. Remember the HDR craze?

I notice incongruencies most often with lighting and colour alterations: while the sky is an early morning or late evening bright orange, the gloss on the bird’s feathers lack that warmth. Nonconformities also show up in overly-sharpened crisp, clear fur and feathers which are far more perfect than one would ever see in nature. I call them hyper-realistic. Another common AI error which is often missed, are shadows in the wrong place or shadows without anything there to cast the shadow.

User Beware

Perhaps the biggest problem with GenAI involves entering photos in contests or competitions. Photographers must read the fine print. You must be aware of which AI tools are ‘internal’ and which are sourcing pixels from elsewhere. Competition organizations, especially those dealing with nature photography, are very strict on what processes can and cannot be used. 

Case in point: the Canadian Association of Photographic Arts (CAPA) has gone to great lengths to create a detailed, 15-page document (LINK) which specifies exactly what tools can and cannot be used within a number of commonly-used editing apps. For example, with Lightroom, you may use the ‘Remove’ tool, but not with ‘Use Generative AI’. With Photoshop, you may use some Neural Filters, but not all, as pointed out above.

But then you may think—how can the judges tell? That depends on three things:

1. how honest the photographer is;
2. how realistically the AI is applied, and
3. how closely the photos are scrutinized.

See this Audubon article about AI in nature photographs, and this CNET article to test your powers of discernment. Nowadays, before winning photographs are announced, each photographer must submit their original JPEG or raw image file for comparison to the submitted image. If the judges are suspicious, they can and will revoke the award, as happened with a Wildlife Photographer of the Year award winner in 2017. (LINK)

“The photographer, Marcio Cabral, denies he faked the scene and claims there is a witness who was with him on the day.” Still, his photo was checked by five independent scientists and all came to the same conclusion: a stuffed anteater from a nearby lodge and the anteater in the photo were one and the same. The award was revoked. In this case it wasn’t AI that was used, but AI can easily be used in the same way.

Authentic Photography

To me, it’s all about authenticity. If any part or pixel of a photograph has been generated using External GenAI, one needs to ask: Is it still a legitimate photograph or should it be deemed something else, such as digital art? CAPA uses  the definition: “a captured image on a light-sensitive device (e.g. film camera, digital camera, smartphone, tablet, etc…) and recorded on film or in a digital format.” (LINK) If the pixels originated from another image, e.g. by using External GenAI, then it is not permitted. Is it still a photograph? Perhaps, but not an ‘authentic photograph’.

A definitive definition of a photograph is difficult at best, and is likely irrelevant. Historically, we’ve accepted when photographers had their prints and negatives  retouched, thereby changing the original capture. Is Lightroom’s Remove tool any different? It is when a GenAI algorithm introduces new pixels from another photographer.

Perhaps we need to become more declarative, as in ‘This is an authentic photograph made entirely by the photographer’ or simply, ‘No part of this image was generated using External GenAI’. This is similar to the growing pressure for clear labelling of ultra-processed food. Perhaps when GenAI is used in a photograph, it needs to be labelled as such, which is beginning to happen, e.g. in Facebook and Instagram.

This may well make room for an ‘Authentic Photography’ movement, similar to the ‘Real Food Movement’. Note: this is different from the Straight-out-of-the-Camera folks (SOOC) who renounce editing of any kind. There needs to be space, a distinction, for those who choose not to use External GenAI in their workflow. Along those lines, Radiant Image Labs has declared that their software, Radiant Photo 2, uses only Assistive AI, not Generative AI and they have committed themselves to authenticity in photography. There is some talk of Serif Affinity going the same route. One may think, ‘That’s economic suicide—everyone is going AI.’ Not true. Niche marketing is alive and well.

Will ‘authentic photography’ become a niche medium, one that may earn a premium? Again, time will tell. Radio was deemed dead once TV became popular, but radio is still with us, and despite the onslaught of digital music, vinyl LPs are still being pressed, and, of course, gelatin silver prints are still made along with palladium prints, cyanotypes, etc.

AI Photography

So, as photographers, where does AI leave us? Will we become flabby and suffer ill effects from Generative AI just like we do with ultra-processed food? Yes. And no. Generative AI is convenient and convenience itself breeds laziness. Many in our society are far less healthy simply due to the convenience of ultra-processed foods combined with the lack of activity spawned by TV remotes, drive-thus, and cars. Reduced health in car-oriented societies is conclusive.

Will photographers will become lazy in their pursuit and execution of photos simply because they can generate what need with GenAI? Consider this: whatever lousy sky they get in a spur-of-the moment shot, can be replaced. Some would argue, ‘Why get up at stupid o’clock in the morning, when I can adjust the lighting and colour effects of what ever photo I take to turn it into a Golden Hour beauty?’ As Luminar Neo tells us, “Twilight photos without waiting for the magic hour.”

As many point out, ‘We’ve been Photoshopping out imperfections for decades—what’s the big deal?’ ‘Besides, who would ever know?’ They have a point. The general public isn’t very discerning and much of commercial photography, especially in media and advertising, is throw-away—used once and gone. It’s the look that counts, not how you produce it. Isn’t it?

AI is here to stay. In fact it’s becoming better and better at creating photo-realistic images. Did you look at those photos from CNET? Already, you can give plain-English descriptive commands to software, such as Midjourney and Adobe Firefly, to generate images that can be further tweaked as you wish. And this is just the beginning. The time is not far off when the result will be fully photo-realistic—client- or printer-ready images at the specified output resolution. Photographers may well be out of a job simply because someone with better language skills will be doing a more efficient and effective job with AI. And, the lighting, the mood, the whole feel of the image can be changed with a few clicks. Art Directors will have all the creative freedom and control they’ve always sought.

It’s all down to economics

As always, the bottom line is economic. In an era of fast fashion and 24-hour news cycles, the convenience of GenAI has the potential for reducing production costs. The economics of GenAI images produced on the spot without the time and cost of hiring photographers and models will make GenAI photos commonplace in media and advertising, a cost-saving that simply can’t and won’t be passed up. Photographers may  still be needed to photograph the item, but after that?

Will AI photography end up like ultra-processed foods—a quick fix for the masses? Definitely, yes. It’s already built into phone cameras. Serious photographers will continue to be more discerning. However, most consumers couldn’t care less if what they’re looking at is real or AI’d. It may sound flippant, but it’s true. And, let’s face it, for many run-of-the-mill commercial images, it wouldn’t matter one way or the other. What ever looks good, right?

But also, no. There will always be room for those who appreciate story-telling by humans, and the art and craft of making fine, authentic photographs. Authentic photography will likely become niche, like vinyl records and gelatin silver prints, but it will still exist.

In the meantime, it comes down to, ‘photographer beware’. Choose and use your tools wisely. Use GenAI all you want, but, if you plan to submit photos for contests or competitions, or simply want to work within the limitations of authentic photography, then be careful of the apps you edit with and the tools you make use of.

Thanks for reading! If you have any questions about AI and Authentic Photography, be sure to add a comment.

---

This work is copyright ©2025 Terry A. McDonald
and may not be reproduced in whole or in part without the written consent of the author.

Please SHARE this with other photographers or with your camera club, and consider subscribing to receive an email notice of new blogs.

Have a look at my work by visiting www.luxBorealis.com and consider booking a presentation or workshop for your Nature or Photo Club or a personal Field & Screen workshop at Workshops.