When Does Photoshop Become AI? Understanding the Line Between Editing and Generating

 Work in progress

I use Photoshop extensively in creating my artwork. I also use Lightroom, Topaz, a Wacom tablet, photography, and traditional artistic techniques. Recently, however, I encountered a question that sounds simple but turns out to be surprisingly complicated:

Do I use artificial intelligence in my artwork?

My first reaction was to think about obvious generative AI, the tools where you type a description and a computer creates an image. But when I started researching Photoshop, I discovered that the line between traditional digital editing, artificial intelligence, and generative artificial intelligence is much less clear.

In fact, many Photoshop tools that photographers and digital artists have used for years incorporate some form of artificial intelligence or machine learning.

AI Does Not Necessarily Mean AI-Generated

This was the most important distinction I discovered.

There are really three categories worth considering:

Traditional digital editing

These are familiar Photoshop techniques such as layers, masks, curves, levels, cropping, color adjustments, blending modes, transformations, brushes, cloning, dodging, and burning. The computer provides the tools, but the artist makes the decisions and manipulates the image.

[EXAMPLE NEEDED: Screenshot of a Photoshop Layers panel showing multiple adjustment layers, masks, and blending layers.]

AI-assisted editing

Here Photoshop analyzes an image and helps perform a task. The software may recognize a person, determine where a sky begins and ends, find an object, repair an area, or make a sophisticated selection.

The computer is helping me accomplish something, but it is not necessarily inventing the artwork.

Generative AI

This is different. Generative AI can actually synthesize new visual information, creating pixels that were not present in the original photograph or artwork.

That distinction becomes very important when someone asks, “Was AI used to create this?”

Photoshop Has Been Using AI Longer Than Many of Us Realized

Consider Select Subject.

Instead of carefully tracing around a person or object, I can ask Photoshop to identify the subject automatically. Adobe describes this as an AI-powered feature using machine learning.

[EXAMPLE NEEDED: Screenshot showing an original photograph beside the result after choosing Select > Subject, with the marching-ants selection visible.]

The artistic subject has not suddenly been created by AI. It was already in my photograph. AI simply helped Photoshop recognize its boundaries.

The same general principle applies to tools such as Object Selection and Remove Background.

[EXAMPLE NEEDED: Screenshot of the Object Selection Tool identifying an object in one of my photographs.]

These are excellent examples of AI assisting the artist rather than replacing the artist.

Then Things Get More Complicated

Content-Aware Fill is particularly interesting.

Photographers have used content-aware technology for years. Suppose there is a distracting telephone pole, garbage can, tourist, electrical wire, or other unwanted element in a photograph. Photoshop can analyze surrounding pixels and construct replacement information.

[EXAMPLE NEEDED: Before-and-after screenshot showing a distracting object removed with Content-Aware Fill.]

Is that artificial intelligence?

Adobe includes Content-Aware Fill among its AI-powered Photoshop capabilities.

But most people would probably view removing a trash can from a photograph quite differently from asking an AI image generator to create an entire Paris street scene.

Technically, both can involve artificial intelligence. Artistically, they represent very different processes.

Generative Fill Crosses Another Line

Photoshop’s Generative Fill makes the distinction much clearer.

I can select an area of an image and ask Photoshop to add something that was not originally there, replace an existing object, or create something that fits the surrounding scene.

[EXAMPLE NEEDED: Screenshot showing the Generative Fill contextual taskbar and prompt field.]

For example, imagine that I photograph a rural Pennsylvania farm. I might traditionally adjust the color, paint into the image, crop it, remove an electrical wire, or enhance the atmosphere.

With Generative Fill, however, I could potentially select an empty field and request a horse.

Photoshop would create a horse that was not in my photograph.

[EXAMPLE NEEDED: Before-and-after example showing an original photograph and a clearly labeled demonstration in which Generative Fill adds an object.]

That is no longer simply identifying or adjusting pixels that already exist. New visual content has been synthesized.

For purposes of describing an artistic process, I consider that generative AI.

Generative Expand Is Another Good Example

Suppose my photograph is horizontal but I need a taller composition.

Traditionally, I might crop differently, clone existing areas, paint additional background, or decide that the photograph simply will not work in that format.

Generative Expand gives me another possibility. Photoshop can extend the canvas and generate plausible visual information beyond the boundaries of my original photograph.

[EXAMPLE NEEDED: Three screenshots showing the original image, expanded canvas, and final Generative Expand result.]

The original photograph may still constitute most of the finished artwork, but some portion of the scene was generated by AI.

That is something I think is reasonable to disclose.

Even the Remove Tool Has a Surprise

One of the most interesting things I discovered involves Photoshop’s Remove Tool.

Removing an unwanted object sounds like ordinary photo retouching. However, current versions of Photoshop can provide options controlling whether generative AI is used during removal.

Depending upon the setting, generative AI may be on, off, or Photoshop may automatically decide whether to use it.

[EXAMPLE NEEDED: Screenshot of Photoshop’s Remove Tool options showing the Generative AI setting.]

That means an artist could potentially use generative AI without thinking of the operation as “generating an AI image.”

This illustrates why a simple yes-or-no question about AI does not always adequately describe a modern digital workflow.

Neural Filters Are AI Too

Photoshop’s Neural Filters provide another category.

These filters use machine learning for sophisticated image processing. Depending upon the filter, Photoshop can perform operations that would once have required considerable manual editing.

[EXAMPLE NEEDED: Screenshot of the Neural Filters workspace showing available filters.]

Again, I think context matters.

Using machine learning to assist with an editing operation is not necessarily equivalent to asking an AI model to invent the subject, composition, lighting, and content of an artwork.

Both involve AI technology, but the artist’s role can be dramatically different.

My Working Definition

After researching this, I have developed a much more useful way of thinking about AI in my own artwork.

I divide the process into three categories:

Traditional digital tools: I manipulate the image.

AI-assisted tools: Photoshop analyzes the image and helps me manipulate it.

Generative AI tools: Photoshop synthesizes visual content that was not originally present.

[EXAMPLE NEEDED: Screenshot or graphic showing these three categories side by side: Traditional Digital Editing → AI-Assisted Editing → Generative AI.]

That distinction is much more meaningful to me than simply asking whether a computer program containing AI was involved.

What About the Artist?

There is another issue that gets lost in discussions about artificial intelligence: authorship involves much more than pixels.

I decide what interests me.

I take the photograph or develop the concept.

I choose the composition.

I determine what stays and what disappears.

I decide on color, texture, mood, and presentation.

I may paint digitally using my Wacom tablet. I may combine photographs. I may use Photoshop, Lightroom, or Topaz. I may work from photographs, sketches, imagined places, or combinations of them.

And, yes, I may sometimes incorporate AI-generated elements into a larger work.

[EXAMPLE NEEDED: Screenshot of one of my complex Photoshop projects with the Layers panel open, illustrating the number of individual components and artistic decisions involved.]

The finished image may therefore be the result of photography, digital painting, compositing, conventional image processing, AI-assisted tools, and sometimes generative technology.

Calling the entire result simply an “AI image” does not adequately describe that process.

Neither, in some circumstances, would saying that “no AI was used.”

Why Disclosure Matters

This became particularly relevant to me because galleries, exhibitions, and organizations increasingly ask artists about AI.

Those questions are legitimate. A gallery has every right to establish rules concerning what it will exhibit.

But artists also need terminology that accurately describes contemporary creative tools.

If I use Photoshop’s AI-powered Select Subject to isolate a person whom I photographed, that is fundamentally different from generating a fictional person.

If I remove a telephone wire using an intelligent editing tool, that is different from generating an entire landscape.

And if I use Generative Fill to create an object that was not present in my photograph, that should be distinguished from both.

[EXAMPLE NEEDED: A three-panel visual using one photograph: original; AI-assisted selection/edit; generative addition. Clearly label each.]

The Question Has Changed

Perhaps the question should not simply be:

“Did you use AI?”

Today, a better question may be:

“How was AI used?”

Was it used to make a selection?

Was it used to remove a distraction?

Was it used to enhance image quality?

Was it used to extend an existing photograph?

Was it used to generate one element?

Or was the entire underlying image generated from a prompt?

Those are very different creative processes.

For me, the goal is not to hide the technology I use. It is to describe it accurately.

Photography itself has always evolved with technology, from film emulsions and darkroom manipulation to digital sensors, Photoshop, computational photography, and now artificial intelligence.

The tools continue to change.

The more interesting question is what the artist chooses to do with them.