The writing world has begun drawing a line.

On one side is work described as human-authored. On the other is work described as AI-assisted or AI-generated.

The language varies. The distinction does not.

For the moment, how a piece of writing was made can determine what kind of writing an institution considers it to be.

There are reasons for that. AI raises questions about authorship, originality, copyright, fairness and disclosure that writing institutions did not have to answer in quite this form before. A new technology has entered an established art, and institutions need somewhere to put the difference.

Film has been here before.

Not with AI. With sound.

The first Academy Awards honored films released in 1927 and 1928, just as synchronized sound was beginning to transform the industry. Wings, a silent film, won the first award that became Best Picture. The following award year, The Broadway Melody, a sound film, won.

The Academy did not create permanent categories for sound pictures and silent pictures. Sound entered movies, while achievement in sound became something the Academy could recognize separately.

Color took another route.

Black-and-white and color cinematography presented sufficiently different problems that, beginning in 1939, the Academy generally judged them separately.

There was Best Cinematography, Black-and-White.

And Best Cinematography, Color.

That sounds, from our side of history, like a transitional arrangement. Color spread. Eventually the categories merged.

But that smooth history leaves out 1957.

For films released that year, the Academy combined black-and-white and color cinematography into a single award.

For 1958 films, it separated them again.

The separate awards remained through 1966. Beginning with films released in 1967, black-and-white and color cinematography once again competed for a single award.

There is something useful in that reversal.

The people living through a technological transition do not necessarily know when the transition is over. Even the institution responsible for drawing the categories can take down a boundary, discover that the distinction still matters, and put the boundary back.

Other distinctions have moved without disappearing.

We still know whether a movie was photographed on film or digitally. Cinematographers care. Filmmakers care. Historians and archivists certainly care. The information remains part of the provenance of the work.

But there is no Best Cinematography on Film and Best Digital Cinematography.

The provenance survived.

The classification did not.

The Academy has now begun making decisions about AI itself.

Under the rules for the 99th Academy Awards, the use of generative AI and other digital tools neither helps nor harms a film’s chances of nomination. In evaluating a film, members are instructed to consider the degree to which a human was at the center of the creative authorship. The Academy also reserves the right to ask for information about the nature of the AI use and the extent of human authorship.

There is no separate AI-assisted film category.

For its Writing awards, however, the Academy draws a harder boundary: the screenplay itself must be human-authored.

Another industry has chosen differently.

In July, recording-industry organizations introduced separate labels for AI-Generated and AI-Assisted recordings. One identifies recordings in which generative AI produced all or the primary portion of the sound recording. The other identifies substantially human-created recordings in which AI was used for some elements of the sound recording.

Two institutions confronting the same technology have already begun drawing different lines.

Visual effects offer another variation.

Computer-generated imagery became part of filmmaking without producing a permanent division between CGI movies and non-CGI movies. The Academy already had a place in which the work could be judged: visual effects.

And Best Visual Effects does not reward the presence of CGI.

Having used CGI is not an achievement.

The quality of the work is.

That distinction changes the question.

A technology can become ordinary while the skill with which it is used remains available for judgment.

It is tempting to imagine the same thing happening with AI and writing.

Perhaps someday there will be recognition for particularly accomplished use of AI in creating a work.

But the analogy begins to break there.

The visual effects in a film can be a separable craft. There are visual-effects supervisors, artists, technicians and companies whose contribution can be identified within the production. The same is true of sound. The Academy can absorb a technology into filmmaking while still handing a statue to the people practicing the specialized craft that grew around it.

That may happen with AI too. New specialties may develop around it. Nobody inside a technological transition should be especially confident about which professions will exist on the other side.

But consider the narrower case of a writer using AI inside composition itself.

The writer asks.

The writer rejects.

The writer keeps.

The writer notices that an answer has flattened the thought and goes back.

A paragraph moves.

A sentence disappears.

An argument that seemed promising turns out not to work.

Something unexpected appears and is followed.

The writer decides when the machine has misunderstood the work and, more dangerously, when it has understood it too easily.

Where, in that process, does the separate craft begin?

If we created an award for excellent use of AI in composition, what exactly would we be rewarding?

Selection?

Judgment?

Structure?

Revision?

Restraint?

The ability to recognize a good sentence and discard a bad one?

Those sound suspiciously like writing.

That does not tell us what will happen to the category AI-assisted.

It may remain useful. It may divide into more precise categories. A separate AI craft may emerge.

1957 should make anyone cautious about announcing that a boundary has reached the end of its useful life.

But there is another possibility, and the current categories make it harder to see.

The disclosure could remain while the classification disappears.

A writer could say that AI was used extensively in making a work without that fact determining the class in which the work is judged.

Nothing would have to be concealed.

The record of making could become more precise, not less. A reader, publisher, archive or awards body could know how AI entered the process, just as we can know whether a film was photographed digitally or on film.

Provenance and classification are different questions.

One asks:

How was this made?

The other asks:

What is it?

At the moment, AI-assisted often answers both.

It may continue to.

But another arrangement is possible. The decisions that make AI useful inside composition may prove difficult to separate from the decisions that make the writing good.

Some technologies become new crafts.

Others enter old crafts and change what mastery of them requires.

WE&P by: EZorrillaMc&Co.