These days, we photographers can’t go two minutes without encountering an opinion about AI. It’s usually negative and often extreme (e.g., “AI will ruin photography forever”), and I can certainly empathize. After all, one of the most well-recognized forms of artificial intelligence in photography—AI image generation—threatens to reduce the value of photography (and photographers) by filling the world with pixel-perfect photo lookalikes that can be produced in an instant and without apparent skill.
But like most things in life, the truth is a lot more complicated. AI image generation gets a lot of airtime, but it’s far from the only instance of AI in photography. Artificial intelligence now appears in a variety of ways and with very different purposes; there are AI photo generators, AI photo organizers, AI photo editors, AI editing tools, AI cameras, and more.
As a result, it feels unfair to talk about “AI in photography” as if it were a single entity. And even if we were to focus on a single AI use case in photography, it likely affects individuals differently; for instance, while many photographers dislike AI image generators, others have enthusiastically taken up image generation as a method of artistic expansion.
Given all the discussion of AI in photography and all of the misunderstandings, I want to spend some time exploring the various ways that artificial intelligence appears in the photography world, as well as the benefits and drawbacks. As a photographer with a longstanding interest in AI—and as someone who has tried a slew of AI-powered photography tools in both a personal and a professional capacity—I’m well-placed to help you understand what AI in photography actually is, how it works, and what it can mean for the world of photography more broadly.
Let’s dive right in!
What Is AI in Photography?
In my experience, most people have misconceptions about the nature of artificial intelligence. The general assumption is that AI is a technology capable of learning from the user—that as the user offers input, the AI gets better at doing what the user wants (in contrast to algorithmic technology, which is static and can only follow an unchanging set of instructions).
But while certain AI software does indeed work this way, it’s generally not the case in photography applications. Technology that learns regularly from user input is often costly since it requires constant updating with each new input. And in certain applications, it produces subpar results because one-off inputs from a single user generally don’t provide enough training data for an AI model to learn effectively.
Instead, most AI in the photography sphere works differently:
First, a model is trained on a huge amount of data to do some specific task (e.g., identify objects in a photo).
Second, the AI model is disconnected from the learning environment and packaged into a photography product.
The result is an AI-powered application capable of impressive feats (e.g., automatically applying keywords to photos) but that is no longer adjusted in response to new input.
A Quick Aside on Artificial Intelligence and Machine Learning
In the photography world, artificial intelligence often gets confused with machine learning. These are two related but distinctly separate concepts.
Machine learning is a process. Through machine learning, models learn patterns from data to make predictions without being explicitly programmed.
Artificial intelligence, on the other hand, is a result. When a program can successfully complete a sophisticated task—such as mimicking an editing style—it’s generally categorized as “AI.” In other words, “AI” isn’t a precise label but is instead a general classifier applied to various software with “intelligent” capabilities.
Additionally, there’s no single underlying method for creating artificial intelligence. While many AI models are indeed produced through machine learning, that isn’t always the case (and not all machine learning results in what would conventionally be called “AI,” either).
Types of AI in Photography
As I emphasized above, artificial intelligence in photography doesn’t appear in just one form. Rather, there are different types of AI photography software, each with its uses, benefits, and drawbacks.
In this section, I walk you through the most common types of AI in photography, starting with:
AI Photo Editing
Over the last few years, AI tools have been incorporated into nearly all mainstream image-editing programs. Skylum led the way with its early AI-powered sliders and one-click sky-replacement tool, but today, you’ll find AI functionality in a wide range of editors, including Lightroom, Photoshop, ON1 Photo RAW, and DxO PhotoLab 8—so if you have access to the latest version of a popular photo editor, then you likely have access to a form of AI.
These AI tools are generally capable of tailoring an editing adjustment or mask to the specific image file. For instance, Lightroom’s AI masking tools are capable of identifying and selecting specific image elements (such as skies) or even facial features (such as eyes):
While Luminar Neo’s Enhance AI sliders apply various aesthetically pleasing adjustments (e.g., increased/decreased contrast and saturation) based on the specific characteristics of the image:
The idea here is to save time by avoiding tedious manual tasks (such as hand-selecting a subject’s facial features), but AI editing tools can also edit images beyond the user’s current skill level to create enhanced results. For instance, a landscape photography beginner may not recognize that their sunset photo would benefit from a contrast boost selectively applied to the clouds, but Luminar’s Sky Enhancer AI slider might do exactly that.
Of course, the level of AI integration varies widely from editor to editor. Skylum’s Luminar Neo includes a slew of AI-powered tools, while Lightroom relies far more heavily on manual features and has added AI editing tools much more incrementally.
AI photo editing exists in other forms, however. Companies such as Neurapix and Imagen offer a more comprehensive, streamlined AI editing workflow; here, the photographer submits photos they’ve already edited in their own style, which are used to develop—through machine learning—an AI “preset” (or “profile”) capable of consistently replicating the style. The preset can then be applied to many photos at once, so when the photographer needs to process a new batch of photos, they can skip manual editing entirely.
Finally, some companies offer software and tools designed to enhance images, not by adjusting editing parameters but by adding or swapping out pixels. Topaz Labs’ Gigapixel software, for example, uses AI to sharpen, denoise, and upscale images, which allows photographers to improve (or even save) files that are soft, noisy, and low quality.
AI Photo Management
AI photo-management tools are often overlooked by photographers, but I think this is a mistake. Here, I’m referring to programs that use AI models to help photographers organize, locate, and/or cull their image files. And while image management may not seem like an exciting topic, these AI tools can save photographers a huge amount of time and effort.
AI image organizers vary in their capabilities, but standard tools include:
- Smart tagging, where the software uses AI to automatically apply keywords to photos.
- Deduplication, where the software uses AI to identify duplicate files in an image catalog.
- Facial-recognition search, where the user can quickly find (and tag) specific people in their photos.
A few programs go far beyond the basic AI photo-management toolkit, however. For instance, in addition to the three tools listed above, Excire Foto 2025 offers:
- Prompt search, which allows the user to retrieve images from anywhere in their catalog by inputting a natural-language description of the photo.
- Smart culling, which groups photos based on a variety of characteristics (including visual similarity, burst sequences, and content) and can even automatically select the best photos from a large batch using criteria such as sharpness and exposure.
- A variety of other search and image sorting functions, including similarity search, face search, and aesthetic scores.
One concern with many AI-powered photo managers is privacy: In order to make images “visible” to AI models, files often get sent off for analysis in the cloud or to a third party. This is legitimate, and it’s one of the reasons why Excire’s software is so popular; the AI functions work locally on the user’s computer so that nothing is sent off for analysis.
Bottom line: If you’re concerned about privacy, just make sure you choose your photo manager wisely. I myself regularly use Excire’s AI photo-management tools to keep my image catalog in order. (Before I adopted AI tools, my photo library was an absolute mess, but thanks to automatic keywording, prompt search, smart culling, and more, my images feel far more organized and accessible!)
AI Image Generation
Thus far, I’ve discussed AI photo editing and AI photo organization, and for the most part, these AI applications work in a straightforward, relatively uncontroversial way: by acting on the photographer’s existing images to find, categorize, mask, and enhance.
Some AI editing tools have drawn criticism for the ease with which they allow for “unnatural” manipulation of an image (Skylum’s sky-replacement tool is one example), while AI sliders and one-click presets occasionally come under fire for simulating artistic processes and decision-making. On the whole, however, these AI applications have been broadly accepted (or at least tolerated) by the photographic community.
AI image generation is a very different story, however.
I’m referring to tools that either:
- Produce photo-realistic images in response to prompts
- Add photo-realistic elements to an existing image
Midjourney, DALL-E, and Stable Diffusion are all well-known examples of the first type of image generator. The user types in a prompt, such as “Photo of a puppy playing in a field,” and the AI model produces a corresponding “image,” as I did using OpenAI’s DALL-E:
The second type of image generator is generally found within more comprehensive editing programs. Photoshop’s Adobe Firefly, for instance, allows the user to input prompts that generate new image elements on top of existing image elements. (Photoshop also offers a Generative Expand feature, which generates new content beyond an image’s original borders.)
These AI tools are controversial. There are many reasons for this, but three come up frequently:
1. Generative AI Causes Confusion
The concern here is that an AI-generated image may be visually indistinguishable from a photograph, yet it’s not tethered to reality.
Therefore, an AI image of people playing football in the park might be persuasive to the viewer (that is, the viewer believes that the football game occurred as depicted), yet the game never happened, and the depicted players never existed.
Such an example may seem harmless—after all, who cares about whether a random football game ever took place?—but for at least some viewers, the reality of an image is important.
Additionally, image generators can be used to persuade viewers of more harmful truths, as concerns around AI deepfakes have shown.
A major caveat here, however, has to do with the quality of the AI-generated images. If AI photos are easy to identify (as many currently are), then this concern is alleviated somewhat—though it’s difficult to imagine that we won’t soon reach a stage where AI-generated files are indeed indistinguishable from conventional photographs.
2. Generative AI Rips Off Photographers‘ Hard Work
This criticism isn’t about the resulting image but about how it’s produced. Image generators are trained to create fake images by analyzing millions of real images, all of which were created by photographers and artists.
There’s a potential legal issue here—how were the training photographs obtained, and did the creator consent?—but there’s also an ethical one: If a model produces “photos” simply by identifying patterns of pixels in existing images, the result is unoriginal and is a form of copying/plagiarism (or so you might argue).
3. Generative AI Threatens or Cheapens True Photography
This concern has several layers, but put simply: If AI can replicate photographs with very little cost and effort, won’t photographers become obsolete? Why would a company buy a stock photo for their marketing brochure or an abstract flower print for their waiting room walls when they can generate a custom image for just a few pennies?
To me, there is at least some truth to this argument, though it doesn’t seem relevant to all forms of photography. AI-generated wedding photos, for instance, will likely remain unpopular, for obvious reasons—though I will note that AI-generated self-portraits have become fairly popular over the last couple of years.
AI Camera Software
Briefly, I want to mention one more AI photography application: camera technology.
For instance, AI is now incorporated directly into some cameras’ autofocusing technology, which allows for better AF performance; Sony’s a7R V camera relies on AI to effectively identify subjects, including eyes, trains, cars, birds, planes, and more.
So far, this AI camera technology has remained relatively underdeveloped, but I expect to see a significant expansion of AI in cameras in the future. I wouldn’t be surprised if cameras five years from now harness AI to meter more effectively, blend exposures at the moment of capture to handle high-dynamic range scenes, optimize white balancing, provide compositional guidance on the fly, apply in-camera perspective correction, and more.
So Is AI in Photography Good or Bad?
At the end of the day, AI is not monolithic; rather, it’s a category of “smart” tools, each with its own possible uses.
And as I hope this article showed, it’s therefore difficult to make blanket statements about the value of artificial intelligence applications in photography. Many AI tools—such as image organizers, editors, and camera technology—do indeed save a lot of time for photographers and offer comparatively few drawbacks. I myself rely heavily on an AI-powered image manager (Excire) and an AI-equipped editor (Lightroom Classic), and I certainly wouldn’t protest if my next camera firmware update included AI-assisted object detection.
On the other hand, AI image generators are more difficult to parse. These programs bring up thorny issues regarding copyright, the nature of art, and the potential to negatively affect photography professionals.
At the end of the day, it’s up to you to decide how you feel about each form of AI. I encourage you to ask your own questions, form your own opinions, and make decisions accordingly. Good luck!
