Artificial Intelligence (AI) has taken the internet by storm in recent years. A variety of technological breakthroughs, including advancements in computing power, hardware, and mathematical modeling, coupled with the availability of extensive, high-quality data sets, have allowed AI to make huge leaps forward. As a result, the topics of AI, its role in our future, and all of its potential benefits and drawbacks, have found their way into countless headlines and debates across the internet and broader media landscape. Given that you’re reading this online, you may have at least a passing familiarity with things like ChatGPT, which has almost certainly generated words that you’ve read on websites, in blog posts, on social media, or even in emails you’ve received.
I’m not going to dive into the intricacies of AI as a technology. I’m far from an expert and there are plenty of resources out there that already provide that information. What I want to touch on is how generative AI may affect photography and, more specifically, nature and landscape photography–both in the near and not-so-near future. I won’t have all of the answers here (or potentially any of them) and, given that the technology is advancing and changing so quickly, the answers I do have may become obsolete soon enough. My goal is for this to be the first in a series as I learn more about the technology and as I observe how it may be affecting photography over time. As opposed to an expert summary on the topic of AI, consider this to be a basis for a larger and ongoing discussion.
What is AI?
While I don’t want to delve too deep into a history and summary of AI, it is important to define what it is we’re covering to provide some guardrails for discussion. AI, as shown in science fiction movies, is essentially robots becoming sentient, rising up, and destroying humanity. But in today’s world, AI is much less apocalyptic, and the Roombas have not yet tried to stage a rebellion. Frankly, until they can avoid being incapacitated by a smartphone charging cable I left on the floor, I’m not very concerned.
Artificial intelligence, in its most general sense, is technology and computer systems that have the ability to perform tasks that are usually associated with human intelligence. Examples of this would include decision-making, pattern recognition, and learning from different situations and experiences. We’re already surrounded by AI that is integrated into our everyday lives. Siri and Alexa are forms of AI. AI is used in banking to detect fraud and in healthcare to identify potential evidence of disease. And, as mentioned above, AI is used to help vacuum some of our homes, even when we’ve left furniture and other non-charging-cable objects on the ground that may need to be avoided.
The advancements in AI that have become so prominent over the last couple of years involve machine learning and generative AI, both of which are subsets of artificial intelligence. Machine learning involves analyzing data to identify patterns and make predictions. Generative AI analyzes available data and learns from it to create new content. Generative AI in particular is what is currently getting the most attention and freaking people out. Or exciting them. Or both.
Generative AI and Landscape Photography
When it comes to landscape photography, the technology that is causing the most discussion is the generative AI that creates new images. These images are created by programs that are trained on metadata, such as keywords and tags, which help the AI to understand what metadata translates to certain aspects of a visual image. The concern among landscape photographers, and photographers in general, is that these images will ultimately replace images captured by humans with cameras, thereby devaluing our artform and, at worst, making it completely obsolete.
There are a variety of popular programs that have the capability of generating images from text, and more are popping up regularly. The ones I see mentioned most often are Midjourney, Dall E-3, Stable Diffusion, and Adobe Firefly, but these are by no means the only options. Adobe Photoshop also now includes a “generative fill” tool, which can be used to replace portions of an image, or expand it with new visuals. In each of these programs, the user describes an image or portion of an image that they want to be generated utilizing keywords, visual styles, and other characteristics, and the AI program produces images to meet that description. Sometimes the results are truly impressive. Sometimes they’re way off. Sometimes they’re visual abominations. What the concerns of this technology boil down to, however, is two-fold:
Can AI-generated images replace nature and landscape photography?
If AI-generated images can rival those taken by photographers, how do people tell the difference?
And since this is the internet, I know I’m supposed to give a simple, black-and-white answer so you can all go back to scrolling other things. But I’m not going to do that, because life is complicated and filled with nuance.
Can AI-Generated Images Replace Landscape Photography?
This is the most common concern I hear in the crowded little niche of landscape photography. Our photos, already seemingly devalued by the fact that accessibility to technological advancements in cameras has helped lead to oversaturation in the industry, may be on the verge of being completely obsolete. The theory goes that AI-generated images will get so consistently good that there won’t be a need to hike 15 miles to capture photos of sunrise at a beautiful location because it’s easier and cheaper to just type a prompt into an AI image generator than taking the time and effort to capture the real thing. To an extent, I think that concern has merit, but not to the point that the ability for photographers to make money is completely wiped away.
To assess the potential effects of AI-generated images, we have to examine why landscape photographers do what they do. For many of us, the joy comes from the process. I’ve had some truly unpleasant experiences while doing landscape photography, but those difficulties have sometimes led to some of my most memorable photos. The satisfaction that comes with knowing how hard I had to work for a given image substantially increases my love of the final product. Further, while the experiences out in the field can sometimes be challenging, the benefits I get from the process of landscape photography is what has kept me engaged with it for over a decade. Those challenges are important for growth, and the failures make the successes all the more meaningful. Photography has driven me to explore both my local and regional areas, as well as travel to places I may not have considered visiting otherwise. It has pushed me to observe the things around me in different ways. The process of landscape photography, for me at least, is meditative, relaxing, and it recharges my batteries. While this may not be true for every photographer, I’ve talked to enough of them at this point to know that I’m far from alone in that experience. And that experience isn’t going to be replaced by typing a prompt into my computer and seeing what pops up on the screen.
Although, I’ll give AI one advantage. Typing a prompt into my computer is a hell of a lot easier on my busted knees than hiking up a mountain with a heavy camera bag.
How Does AI-Generated Imagery Affect the Business Side of Landscape Photography?
Although it’s fairly clear to me that AI-generated images can’t replace the experience and process of landscape photography that has value to the photographers themselves, that’s also not what most people are concerned about. With the advent of new technologies, the primary concern for many is that it could take money out of our pockets and ruin our livelihoods. I’ve talked to photographers who are concerned about making less money at art fairs or from content creation. I’ve also talked to those who have just a general sense of dread as to what AI means for being able to tell what’s real and what’s fake about a given image. In most cases, the commercial viability of photography is going to come down to the photographer being able to articulate the value of that photography to a customer or client. In some cases, however, even in the near–future there are areas where that very well may be difficult, if not impossible.
Print Sales
Although I’ve made money over the years from writing for websites, teaching, and doing contracted shoots, print sales have made up the most consistent part of my photography income over the long-term. Those print sales typically come from fine art fairs, sales from my website or Etsy shop, and sales from retail locations that carry my photos.
When it comes to selling prints, this is the area that I think will be least affected by the onset of AI-generated imagery. This is also good news because these types of print sales are likely what account for a large portion of many landscape photographers’ incomes. Digital imagery isn’t new. We’ve been able to create digital art or wildly alter images in programs like Photoshop for many years. Yes, AI image generators like Midjourney and Dall-E 3 now make that process easier and, as a result, could flood the market with more high quality images, but that doesn’t change the fact that print sales and sales of art in general are typically tied to the buyer’s connection with and perceived value of a photo.
For me, photography is more than just the final image, but I understand that not many people will view the final product with the same amount or type of value that I do. I may care that I sat in summer holiday traffic for 4 hours to get a photo of a full moon over a lighthouse, but that likely doesn’t increase the photo’s value to a buyer. They care about the image because it is a beautiful scene that represents their hometown or because it elicits fond memories of when they watched the moon rise over that same lighthouse. They care because it captures something real that connects with something real that they experienced. For the vast majority of print buyers, if I told them that the image they were looking at didn’t actually happen and was created by AI as a response to me typing “full moon over a lighthouse,” their connection with the image and their desire to purchase a print of it would likely vanish. If the buyer’s goal is strictly putting a visual they like on a wall with no other meaning tied to it, then AI could very well cut into sales, but for me at least, that wouldn’t affect the vast majority of mine.
Maybe that sentiment will change as AI-generated imagery becomes the rule rather than the exception, but I don’t think those days are close. Until then, I think that print sales will still be a viable way for landscape photographers to make money, but it will remain up to us to articulate the value of our photography and educate our customers. What if other sellers lie about or omit the fact that their images are AI-generated? Don’t worry, we’ll get to that.
Stock Photography and Image Licensing
Unlike print sales, income that originates from stock photography royalties and image licensing has the potential to be heavily affected by AI-generated images. Business, in a nutshell, is solving a problem or filling a need for a client. If that client’s need can be filled by something quick and cheap, like generating an image they need through AI or licensing a stock image that may have been easily created by AI, the forces of supply and demand are going to take their toll on photographers.
To be fair, I doubt stock photography is a major source of revenue for any landscape photographers at this point. If an advertising manager needs a non-specific image of a mountain at sunset for a project they’re working on, there are plenty of services that can supply them with countless results for the cost of a few dollars at most. Personally, I’ve never even made an attempt at submitting images to stock images websites. The chances that someone steals my photo and uses it without permission for a commercial purpose is probably higher than successfully licensing it through a stock website. And if I had licensed that image through one or more stock image sites, it’s going to be difficult to know if my photos were stolen or legally licensed when I come across them online.
For image licensing outside of stock photography websites, there may be a way for AI-generated images and photographs to exist together, but this will ultimately be determined more by market factors like the attitude of businesses and the general public towards AI images than anything that photographers can easily control. If a client needs a specific image of a real-life scene, it’s unlikely that AI–at least at this point–can provide that. From playing around with image generators like Dall E-3 and Stable Diffusion, I found it extremely difficult, if not impossible, to create an image of a real location. The best I was able to come up with was a dreamy image inspired by the characteristics of a real place. If you’re looking to license an image of the Milky Way rising over Skull Rock in Joshua Tree National Park, AI image generation, at this point, isn’t going to cut it. In the examples above, you can see that AI was able to make an image that includes elements of Joshua Tree–its namesake trees and monzogranite rock formations–but failed miserably at recreating the iconic Skull Rock. Likewise, AI made an image loosely resembling Wizard Island and Crater Lake at twilight, but placed side-by-side with the real thing, it leaves a lot to be desired.
Ultimately, the driving force in whether licensing landscape photography will be a viable source of income will depend on two things:
1.) The ability of AI to reliably create images of real places.
2.) Whether clients will continue to desire documentary images of real places versus digital recreations of those places.
Contracted Shoots
Similar to image licensing, whether contracted shoots are affected by AI-generated images will depend on market factors and the general public’s acceptance of AI images. If a client in need of images can generate those images with AI instead of hiring a photographer to go and capture them in the field, they’re going to save the money and do so. However, while opportunities for contracted shoots for landscape photography may not seem to be plentiful, they do exist, and many organizations will likely still require images documenting real places and conditions as opposed to images that just resemble those places. A wilderness conservation organization is going to need actual photos of the places they are trying to conserve. Magazines looking to document actual moments at a real location are still going to need a person to do the documenting. These jobs likely won’t go away anytime soon.
Content Creation
Although I’m largely not in this category, content creation is a way that many creatives, including landscape photographers, try to help make an income. For the sake of discussion, let’s lump writing, social media, and video into this category. The rise of generative AI has probably had the most significant impact on content creation, as many types of it can now either be replicated or replaced to some degree. You’ve almost certainly seen emails, blog posts, website copy, social media posts, and even social media comments written with things like ChatGPT. In some cases, I’m fairly sure I’ve seen back-and-forth discussions in the replies of social media posts entirely comprised of AI-generated responses. The day I realized that was a day I got sad and gave my Roomba a bit of undeserved side-eye.
For content creators, AI may be as much of a help as it is a hindrance. While it may negate the need for publications to ask for us to write an article, creatives like landscape photographers will also be able to benefit from those efficiencies. If you hate posting on social media but feel it’s a necessary part of getting your work out there, you can ask ChatGPT to write a social media post for you. If you never get around to writing blog posts but would still like to have them on your website, AI may be better than just going without them altogether. As much as AI may reduce the need for others to ask us to create content, it could significantly increase the efficiency of creating content that we make for our own purposes.
For reference, none of this post was written by AI. I enjoy writing and storytelling, and I take pride in it, so to outsource my writing to ChatGPT just isn’t something I’m interested in. Also, the irony of using AI to write about AI replacing landscape photography is way more than I’d be able to stomach anyway…
Content Provenance - Knowing What’s Real and What Isn’t
With everything we’ve covered so far, the major issue that arises and permeates all of these different discussions is being able to know which images that we see are real and which are generated with AI. In a perfect world, everyone would be forthcoming with this information when necessary, but if you’ve been on the internet at all in the past 10 years, you know as well as I do that internet content and its authenticity are not two topics that reliably go hand in hand. Because of this, content provenance is something that is going to be more important than ever.
Digital content provenance isn’t as simple as a file showing copyright information. The purpose of the technology is to provide not only the origin story of the image file, but also all of the ways it has been altered before it reaches your eyes. For example, an ideal use for content provenance would be viewing an image on social media, and being able to click on an information icon that shows when it was captured and if it’s been changed in a way that may skew the reality of what you see in the image. Keep in mind that skewing the reality of what is in the image is not necessarily an issue. If you’re simply viewing a piece of art, then those changes are certainly fair game. However, if the image purports to be documentary or photojournalistic in nature, then how that image file was altered after capture (or if it was even captured with a camera at all) could be the difference between assuring its authenticity or telling the viewer that they may not be able to believe what they are seeing.
Progress is already being made on this front by several different groups. The Content Authenticity Initiative (CAI) is an Adobe-led initiative which “focuses on systems to provide context and history for digital media.” Project Origin, which is led by Microsoft and the BBC, focuses on preventing disinformation within digital news. The Coalition for Content Provenance and Authenticity (C2PA) unifies the efforts across these groups. Collaborators include The New York Times, Qualcomm, Truepic, Intel, Arm, Nikon, and more, showing that the efforts on the topic are being made by some of the companies that would be heavily affected by the need for content provenance. Further, Leica has already introduced the M11-P to market, a camera that can apply a cryptographic signature to an image file when captured, and Sony has made their similar In-Camera Authentication Technology available on the a7 IV cameras. Likewise, Nikon has already demonstrated a Z9 camera capable of complying with CAI standards.
Benefits and Drawbacks of Content Provenance
The obvious upside to content provenance is to battle fake images and misinformation. However, as important as this is and will continue to be in areas such as photojournalism, the story isn’t as cut and dry when it comes to landscapes.
In my experience, viewers of landscape photography would change their opinions of an image if they knew it was AI-generated. So, to have technology that separates the verifiably “real” images from the AI-generated one could be an important benefit to landscape photographers. But there has always been a blurry line that divides “real” and “fake” in digital photography. I’ve long been open about how and to what extent I’m willing to process my images, but my chosen limits with digital manipulation aren’t a measuring stick for other photographers.
Most landscape photographers aren’t necessarily looking to show images that are documentary. While content provenance would be beneficial in showing when an image was captured by a photographer and when it was generated by AI, detailed information in content provenance could describe edits made in a program such as Photoshop that a photographer might not want revealed. That desire could be to protect a “secret sauce” technique used in editing to create a particular look or style. It could also be to remain vague about how realistic an image is for fear of potential customers changing their opinion of the image. There are plenty of examples of landscape photographers either lying about the stories behind their photos, or at the very least doing their best to mislead viewers into thinking that images aren’t composite shots of various scenes from various places.
Closing Thoughts
It remains to be seen how generative AI will grow and how the general public will receive it. It will continue to be refined and trained on more content, at times on our own images, which provides not only competition issues, but legal and copyright issues. If AI-generated images become accepted and commonplace, some facets of landscape photography could very well be devalued.
However, just because AI becomes more accessible doesn’t mean that people will lose their humanity, or their desire to connect with humans and the things directly created by humans. Photography didn’t make painting extinct, so I don’t expect that AI-generated images will cause the disappearance of photography. It will just be up to photographers to articulate the value of our photos or the connection to the natural world they provide that generative AI does not.
As I said above, this isn’t meant to provide all of the answers, only to give a basis for discussion. AI is here to stay, but it remains to be seen how recent advancements and future advancements will alter the landscape of photography.
b
Note: Some of the links on this site are affiliate links. If you click on any of these links and buy a product (any product, camera-related or otherwise), I receive a small commission I can later use to further support both my photography and charity, and it costs you nothing extra. Please consider using these links, especially if you would like to support my photography!