The AI Story: How to Collaborate with Machines
Carloscastilla / Alamy Stock Photo
Is AI a narrative agent like any other?
Writing tools powered by artificial intelligence cannot, by definition, be considered as narrating agents in their own right. This is because AI algorithms are created with the aim of matching the skills of a unique type of intelligence, which their designers claim to be identical to human intelligence (without ever defining it, by the way).
A narrating agent is not “anyone”. It is the embodiment of a specific point of view, i.e. a personal way of thinking, subject to subjectivity, intuition, irrationality and various forms of flawed knowledge, moral values and sensitivity.
The text produced by AI does not come from a narrating agent, nor does it constitute a proper narrative. It is the compilation of fragments from different narratives (those which have been used to “train” the machine). The AI cannot therefore “write a screenplay”, which is a central component of the narrative represented by a film, television or comic book; at most, it can contribute to the creation of the plot.
What does a
narrating agent
want?
Three examples of narrative “intelligence” traits that AI-powered writing aids lack :
1. AI doesn’t understand that the plot is not the ultimate goal of a narrative work.
Plot events are filtered through the prism of the narrating agent. There are a multitude of possible narratives for the same plot. Narrative choices are based on criteria that are more or less conscious, more or less motivated, and always potentially contestable.
A plot without a singular point of view turns the audience into a passive voyeuristic entity. A proper narrative includes the receiving agent in a dialectical and emotional construction. The interest of a narrative lies as much in what the narrating agent knows about the plot as in what they are unaware of or cannot/will not see and hear.
2. AI is incapable of handling characters at odds with themselves.
When a self-driving car encounters a road closed for roadworks, it stops and stands still. Artificial intelligence is incapable of reproducing common sense, intuition, projection into the unknown, ingenuity.
Algorithms assume that characters pursue a single objective at a time, whereas an objective may be a means to an end, or (temporarily) in total contradiction with the protagonist’s interests.
What does a
character
want?
3. AI does not ask itself what makes a work of art unique.
The machine bases its criteria on ideas already expressed by others, whereas the value of a work of art is measured by its novelty, its untried, untested character.
AI aims for knowledge (too often confused with intelligence), while the human being explores, through art, what lies behind knowledge, the essence of truth, the meaning of consciousness and existence.
So what use is artificial intelligence in storytelling?
It can be used to create an infinite number of unexpected, unpredictable and mysterious situations. AI can be a surprise machine, as a film director who has just tested it said :
Well, that’s interesting, it’s suggesting I go somewhere I wouldn’t have thought of. That’s what amuses me, like a situationist drift, the machine will take you to something completely unexpected every time.
BOUISSON (Simon) [in] L’ADN, Une AI capable d’écrire des scénarios? (ladn.eu)
But surprise only makes sense if it allows us to discover the reasons why we are surprised. Some surprises are trivial, while others have far-reaching consequences. Only a narrating agent “inhabited” by a conscience can chose the “right” surprises.
What is
Surprise ?
What positive consequences can the advent of AI have on narrative creation?
The ‘natural’ narrating agent can distinguish themselves from the rise of AI by developing their ability to make their distinctly human contribution more perceptive.
The choice of narrative tools, their application, the clues left by intentions, motivations and reticence in the narrative are all ways of penetrating the singular, conscious and unconscious inner world of the main narrating agent, whom I call on this site: the “implicit author”*.
A quality narrative is also a story, the story of the narrating agent. A meaningful narrative not only tells us a plot, but also the story of who is behind it. The narrative experience of the receiving agent (reader, viewer, listener) is that of a transaction between their own story and that of the narrating agent.
The ascension of AI thus should encourage creators of narrative to (once again) confront a few fundamental questions:
-
Why does the plot need to be told?
-
What motivations lie behind the choice of obstacles and narrative factors?
-
What does the narrating agent really want?
-
What is the purpose of each description?
-
Why is the camera looking in this particular direction?
-
What is the meaning of each event?
-
What does the actor have to “say” about the character?
Etc.

