A Work’s ‘Context Capacity’: How Differences Between Media May Shape Creation in the AI Era

How will AI change creative work?

When people discuss this question, paintings, novels, manga, and other forms are often grouped together as “creation.” I think each medium will be affected differently.

One idea that interests me when considering those differences is how much context a single work can contain.

What has a character experienced? What reason lies behind an action? How does a later development change the way an earlier scene appears? How much accumulated background, relationship, and meaning can be woven into a work?

In this article, I will use “context capacity” as a provisional name for the room a work has to accumulate background and relationships within itself. It is not a strict numerical measure, but a tool for thinking about differences between media. Context outside a work, and the amount an audience actually receives, are related to this capacity but are separate questions.

What do we see when we look at a famous painting?

A painting is not appreciated only for what appears on its surface.

When was it painted? What was new in that period? What kind of life did the artist lead, and under what circumstances was the work made? Understanding a work through such background is one of the basic approaches of art history. Introduction to art-historical analysis

Information outside the work also affects its value as an object. Authorship, date, and provenance—the history of who owned it—can influence monetary value. The Getty’s explanation of provenance research

Nor were all works now considered famous received in the same way when they first appeared.

In Van Gogh’s case, efforts to introduce his work more widely continued after his death, while publication of his letters also made his life and ideas better known. He did receive favorable criticism during his lifetime, so the familiar claim that nobody recognized him is inaccurate. Even so, the process by which his work gained acceptance was connected with the spread of understanding about the artist himself. Van Gogh Museum research journal

Context outside a work therefore supports both appreciation and evaluation.

What I want to consider from this is whether different media vary in how much of that context they can place inside the work itself.

Where can context be placed within a work?

Consider a single image and a novel.

A single image can contain many meanings through composition, expression, color, background, and symbolism. But sharing specifically why a person has that expression, or what happened before that moment, requires particular devices. An explanation outside the work is often needed when the subject is the artist’s life or the circumstances of production.

A novel can accumulate that background within the work, in sequence.

A character first appears cold, but knowledge of their past changes the meaning of the same behavior. A casual conversation later proves important. An object that first seemed ordinary becomes a symbol of a relationship between characters.

As readers continue, they receive the context needed to understand the work.

I think a single image and a novel differ greatly in the room they provide for this kind of accumulation. Manga, film, and games each have their own structures for building meaning over time.

That is what I mean by a difference in “context capacity.”

A novel is also affected by the author’s life and historical period. Even so, a reader who knows nothing about the author can acquire substantial background by reading the work and reach the point of thinking, “I like this.” That structure is what interests me.

The objection that images can also contain abundant information

When I raised this idea with AI, it offered an objection.

Perhaps a single image does not contain little information. Perhaps its information is compressed, and the audience unfolds a story from it. Being viewed for a short time does not make it less valuable.

I think that objection is correct.

A person can draw a great deal from one image and spend a long time looking at it. Some pictures become unforgettable after a single encounter.

But what concerns me is the difference between what it is possible to read from a work and what many people actually read from it.

It is common to see an image in a social-media feed, think “That is beautiful” or tap “like,” and move on. How many people then investigate the artist, learn the circumstances in which it was made, and come to understand the creator’s sensibility?

Even when a work contains rich context, that richness does not necessarily become a reason for choosing it if the audience never reaches it.

I suspect this may make earning a living from single images more difficult.

When beautiful work becomes abundant, what makes one work chosen?

Suppose AI produces an even greater quantity of images that look beautiful at a glance. What happens then?

A viewer may see an image carefully conceived and painted by a person and an AI-generated image, respond “Beautiful” to both, and continue scrolling. If the difference is not received at that moment, it becomes difficult for the effort and thought invested in production to be the reason one is chosen.

Relationships of the form “I want this because this person made it” certainly exist. Watching someone’s process can make us want to support them, and admirers may want an original work by a favorite artist.

But before that relationship can form, the work must first be seen.

The sequence often begins with seeing a work, liking it, and then becoming interested in its maker. For someone who does not yet know the artist, how much distinctive appeal can the work itself communicate?

With a novel, choices made by characters, relationships, and accumulated setup provide room for differences to become clearer as someone continues reading. I think creators have room to express their thought and sensibility through those decisions.

This does not mean that human authorship automatically produces distinction. The opportunity for distinction lies in repeatedly deciding what to accumulate, when to change its meaning, and how the reader should receive it.

That is why discussions of AI’s effect on creation may need to consider context capacity, not only appearance or the quality of one isolated part.

Capacity is not enough unless someone receives it

This does not mean novels are safe.

A novel must first persuade someone to begin reading. A long work may contain rich context, but if the reader stops halfway, what lies beyond that point does not reach them. Length alone does not make a work rich.

A single image has the power to attract someone instantly. If a series of images is considered one work, context extending beyond any individual image can grow within it. The unit of comparison changes depending on whether we compare one image with one novel, or a series with a long-form work.

Whether AI is used and whether a work can contain context are also separate questions. A work made with AI, through human thought, selection, and revision, should still be able to reflect that person’s context.

What I want to consider is how much background and meaning each medium can hold within itself, and how much of that can reach its audience.

As AI makes it easier to produce polished things, helping someone feel why this particular work matters becomes more important. Is that reason accumulated inside the work, or does it also extend outward into the creator and the process of making it?

“Context capacity” is a term for thinking about creation in the AI era through that difference.

---

The dialogue behind this article

The following excerpts have been organized from the conversation that led to the article. Repetitions and corrections caused by voice input have been removed from my remarks, and some wording and nuance were adjusted for publication in accordance with my intent. They are not verbatim quotations. The AI’s replies are also excerpted where needed.

Me, on background outside a painting

I think this is common: we appreciate paintings with their historical background and the artist’s life in mind. Along with the painting itself, that background affects its cultural value and its value as a commodity.

Me, on the difference between a single image and a novel

A picture can hold less context inside the work, so more of its context ends up outside it. That outside context becomes the main part of the story.

A novel contains an enormous amount of information. Structurally, an author can put a great deal of context into it, while that is more difficult with a picture.

When people evaluate a work, before they begin with the person who made it, the first question is whether the work itself is good.

The AI’s objection

The largest objection is that a single image does not contain “less information”; it is a medium with high information density that the audience decompresses. A novel explicitly supplies a large amount of information, while an image can let viewers construct a story from composition, gaze, color, clothing, background, symbols, poses, light, and the distance between characters. It may look smaller when converted into a word count, but the amount an audience can expand from one image is large.

A second point is that a short viewing time does not necessarily mean low value.

What I emphasized: how works are actually viewed

I do not think most people spend that long looking at a single image. Usually it is “Nice. Next.”

There are already any number of pictures that look good at a glance, so it is very difficult to distinguish yourself there.

I recorded how I restated my view after hearing the objection in “Ask AI to Argue Back—But Do Not Let a Correct Argument Choose Your Taste” (coming soon). Alongside differences in capacity, I want to keep thinking about whether that capacity actually reaches an audience.