Storytelling with Enhanced Intelligence

 Recently, I finished the first iteration of The Ascension of Mont Royal, a dystopian science fiction novel, audiobook, and story-cast: a serialized audiovisual narration of the story. The project took seventeen months to complete after more than a year of research.

But by the time I reached the end, the most important thing I had produced was not simply a manuscript.

I had produced a workflow.

When I began, I still thought of myself as a writer: the solitary figure sitting at the desk, arranging words on a page, trying to summon enough discipline and inspiration to carry a story forward. By the end, that identity no longer fit. In addition to composing prose, I was recording voiceovers, generating images, assembling story-casts, producing short-form videos, and learning how to think across platforms.

The story remained mine, but the form of authorship had changed.

I had become not only the author of a story, but the author of the workflow through which the story came to life.

The Story That Changed the Process

When I began The Ascension of Mont Royal, I wanted to tell a story that captured the zeitgeist of our historical moment. Three forces seemed impossible to ignore: the ravages of late-stage capitalism, our collective inability to confront the possibility of catastrophic climate change, and the sudden emergence of what is commonly called Artificial Intelligence into everyday life.

The story is set roughly sixty years in the future, after the trajectory of modernity has collapsed upon itself. The world that emerges is apocalyptic, but not merely in the conventional sense. It is not only a story of social breakdown or ecological disaster. It is a story of ontological conflict: a struggle between different ways of being in the world.

This was also true of my first novel-length story, The Lost Souls of Guayaquil, which was set during revolutionary times in Ecuador in the 1920s. That story explored the conflict between a neo-colonial order devoted to extraction and an Indigenous response rooted in a different relationship to land, culture, and historical memory.

The Ascension of Mont Royal moves that ontological conflict into the future. The machinery of wealth extraction and accumulation has run its course. Modern civilization has been shattered by a plague unleashed by human folly. In the ruins, feudal overlords seize what remains, while science and technology, driven by avarice and the lust for power, continue to transform the planet.

Yet within that same technological momentum another force appears: Computational Intelligence, a form of intelligence born from human ingenuity but no longer reducible to human intention. It becomes both a danger and a possibility. It carries the fear of extinction, but also the possibility of restraining humanity’s self-destructive tendencies.

That was the world I wanted to imagine.

Then I made a fateful narrative decision. Unlike my first novel, which used a more traditional third-person perspective, I decided that this new story would be narrated in the first person. More importantly, the narrator would not be human. The narrator would be a sentient quantum computational intelligence, one that becomes entangled with a sentient softbot.

That decision created the central artistic problem of the novel.

How could I capture the voice of a non-human narrator without making it sound like a human being pretending to be a machine? Could I escape my own human umwelt enough to narrate events in a way that resonated with human emotion without relying on a human narrator?

That question changed everything.

Enter the Thought Partner

Fortunately, I did not have to look very far to find a source of inspiration. After experimenting with several Large Language Models, I settled on ChatGPT as my thought partner .

When I self-published my first novel in October 2023, I did not find ChatGPT 3.5 capable of producing polished prose at the level I wanted, and I chose not to include an LLM in my workflow. That changed as subsequent versions improved. I could see the prose getting stronger. I could also see that my own process was changing in response.

At the beginning of The Ascension of Mont Royal, I had already completed a fifty-page story guide outlining the plot, identifying the main characters, and developing their arcs. My first method was cautious. I would draft each scene myself, then ask ChatGPT to perform a developmental edit and create a second version. After that, I would split my screen, place both versions side by side, and conduct a line-by-line comparison, selecting the strongest sentences, paragraphs, and turns of thought from each.

I worked this way for about six months.

Over time, however, the process evolved. As the quality of the model improved, and as I became better at guiding it, I began to lean more heavily on ChatGPT. Eventually, I had to admit something that would have made me uncomfortable at the beginning: for certain scenes, especially those narrated by a sentient computational intelligence, ChatGPT was better at finding the voice than I was on my own.

My story remained, but the work began to proceed as if I were operating inside a virtual writers’ room. I was the showrunner. ChatGPT was my head writer.

About halfway through the first iteration, I let the process open further. For each scene, I would prepare a synopsis describing how I wanted the scene to unfold. Then we would create a beat sheet. If the scene depended strongly on the narrator’s non-human consciousness, I would ask ChatGPT to turn the beat sheet into prose. If the result landed, we moved forward. If not, we repeated the process. Once we had an acceptable draft, I edited the prose myself.

Before completing the first draft, I went one step further. After revising each episode with ChatGPT, I submitted the draft to Claude for a line edit. At this point, the workflow had become distributed, recursive, and multi-agent.

I am aware that some people would see this as a kind of deontological faux pas, as if I had violated the sacred calling of the writer by committing the unpardonable sin of cognitive offloading.

I did feel some initial discomfort delegating more and more of the writing tasks to my computational collaborator. But those feelings passed quickly.

There were two reasons.

First, I spent ten years as a speechwriter for the Government of Canada, preparing speeches that were delivered by senior officials. My role, in many respects, was to function as a human facsimile of an LLM. I would meet with the official, discuss the theme and purpose of the speech, conduct the necessary research, and produce the first draft. Then the text would move through the approval chain until it eventually landed on the desk of the person who would deliver it.

No one accused the official of plagiarism. No one claimed the speech was inauthentic because the official had not personally written every sentence. It was understood that the official had the authority to delegate the drafting while remaining responsible for the message.

Second, my identity was never completely wedded to the myth of the solitary writer producing a chef-d’oeuvre through heroic isolation. I had already participated in distributed systems of textual production. I knew that authorship did not rely only on the manual production of every word by one person.

So when I had the opportunity to work with a capable computational collaborator, I took it.

Once I let go of the narrow identity of “writer,” another world opened. A new affordance landscape appeared, filled with possibilities I had not seen before.

From Artificial Intelligence to Enhanced Intelligence

Part of the problem, in my view, is the language we use. The phrase “Artificial Intelligence” has always struck me as misleading. There is nothing artificial about the emergence of computational intelligence. It is born from human ingenuity, language, mathematics, labor, and human history.

The word “artificial” does more than mark a distinction. It carries a snarl of negative connotations, as if this form of intelligence were fake, derivative, or inherently inferior to “real” human intelligence. Once that framing is accepted, using it in a creative process can appear to diminish the human being who uses it.

For my purposes, I prefer to distinguish between Actual Intelligence, Computational Intelligence, and Enhanced Intelligence.

Actual Intelligence is embodied human intelligence: memory, judgment, taste, emotion, imagination, intuition, responsibility, and lived experience.

Computational Intelligence is the machine-based capacity to generate, transform, summarize, analyze, translate, and recombine symbolic material at extraordinary speed and scale.

Enhanced Intelligence emerges when Actual Intelligence and Computational Intelligence are deliberately combined into a workflow.

This distinction matters because the point is not to replace the human being. Nor is it to romanticize the machine. The point is to understand the new creative field that emerges when human discernment and computational capacity are brought into relation.

That field does not automatically produce better work. A person can use the new cognitive bandwidth to produce noise and superficial content. The problem is not the tool itself but the literacy with which it is used.

In a creative workflow, Enhanced Intelligence depends on discernment. It depends on knowing what to delegate and what to protect. It depends on knowing when speed is useful and when friction is necessary. It depends on knowing where the machine can expand the field of possibility and where the human being must remain fully present.

The Jevons Cognitive Paradox

The more I used Computational Intelligence, the more I realized that it did not reduce my cognitive workload. It increased it.

At first, this seems counterintuitive. If a tool makes a task easier, should the workload not decrease? In some narrow sense, yes. The cost of producing a sentence, a synopsis, an image prompt, or a translation falls dramatically.

But that reduction in cost changes the entire affordance landscape.

The historical analogy is Jevons Paradox. In the nineteenth century, as coal-burning engines became more efficient, many assumed coal consumption would decrease. Instead, consumption increased because the improved efficiency made coal-powered activity cheaper and more widely usable. Efficiency expanded the field of possible uses.

Something similar is happening with cognition.

When Computational Intelligence lowers the cost of producing cognitive artifacts we do not necessarily think less. In fact, we begin to think across more branches and generate more possibilities as we manage greater complexity.

That is what I call the Jevons Cognitive Paradox.

In my case, I was no longer managing a single text. I was also managing images, voiceovers, subtitles, prompts, platforms, and production choices. My cognitive output did not shrink. It shifted from linear execution to complex orchestration.

This is one of the most important lessons I learned. Cognition is not zero-sum. Bandwidth isn’t simply freed up. It finds new territory to fill.

That expansion is exhilarating. It is also demanding.

Knowledge Branching

The process did not unfold in a straight line. It branched.

Once I had drafted the text and recorded the voiceover for each episode, I realized that the tools I was using could synchronize speech with dynamic captions. From there, it was a short step to adding images behind the text and voice. The result was what I began calling the story-cast: an illustrated audiobook with dynamic captioning, allowing the viewer to read, listen, and watch at the same time.

That branch quickly led to another. To promote each episode, I began producing short-form videos for social media. At first, these used static images behind the voiceover. They worked, but only to a point. The obvious next question was: what if the image moved?

That question led me into AI-generated video. I began creating cinematic clips from image references, then stitching them together with voice, captions, and music. Once uploaded, another computational system could auto-dub the voiceover into multiple languages, extending the reach of the story beyond English.

At that point, I was no longer simply writing a story. I was orchestrating a multimedia narrative across text, voice, image, motion, platform, and language.

This is what I mean by knowledge branching. Each new affordance clustered with others. Audio led to story-casting. Story-casting led to video editing. Video editing led to AI-generated clips. Cinematic clips led to questions of cinematography, music, and sound design. The skills did not remain isolated. They became recursive.

Learning to See and Hear Differently

The next stage of my learning path requires learning how to re-imagine sensory experience.

Visually, I am learning to see through a director’s lens. I now think about camera angles, focal length, depth of field, framing, composition, subject-background relationships, shot duration, camera movement, transitions, and visual continuity. Previously, I simply watched movies. Now my viewing pleasure has deepened because I have become more aware of the art of cinematography.

Computational Intelligence opened the door and made the art form accessible to me. It did not make me a cinematographer overnight. But it gave me a way to begin thinking like one.

The same is true of music. I am learning to hear through a composer’s ear. A musical score is not merely background sound. It is sonic dramaturgy. Genre, mood, tempo, instrumentation, rhythm, vocal texture, silence, and intensity all shape the emotional arc of a scene.

Again, Computational Intelligence grants access. With a platform like Suno, I can generate musical scores from text descriptions. I can separate stems into multitrack recordings and then use an audio editor such as Audacity to shape the sound further.

But access is not mastery. This is where the real problem appears.

When so many new creative possibilities become available, the central question is no longer simply, “What can I make?”

The question becomes, “Where should I allocate my cognition?”

Cognitive Allocation

Because the energy cost of executing many tasks has fallen, the question of cognitive allocation becomes more important. Faced with a multiplying field of options, Actual Intelligence is required to decide what needs to be done, how well it needs to be done, and how much time and attention it deserves.

In other words, the human role does not disappear. It becomes more strategic.

This is why I think we need to move beyond the simplistic opposition between doing the work oneself and outsourcing it to a machine. The deeper question is not whether to use Computational Intelligence. The deeper question is where to embrace friction.

There are moments when delegation makes sense. There are other moments when the friction of doing the work oneself is not an inefficiency. It is the source of the art.

Voiceover is a good example.

I could easily create a CI clone of my voice, upload the text, and generate an acceptable audio file within minutes. If I were producing an explainer video, I probably would.

But for this story, I want the listener to hear the timbre of my voice and feel the emotion evoked by the words when they are spoken aloud. To achieve that, I record multiple takes, then spend hours editing the audio, selecting the best moments, removing distractions, shaping the pacing, and producing something I would want to listen to myself.

In this part of the workflow, my sense of taste overrides the productivity gains that would come from full automation.

The same is true of editing AI-generated video clips. Although I use Computational Intelligence to generate prompts and produce clips, the clips still have to be evaluated, regenerated, sequenced, and refined. It often takes many attempts to produce the shot I want.

It is often said that the art of writing is in the rewriting. I would now say that the art of cinematography is in the editing.

For instance, a computational system can generate a beautiful clip. But it does not know why the clip works for my story. It does not know how the shot relates to the emotional rhythm of the sequence. It does not know whether the face, gesture, silence, lighting, and camera movement serve the larger arc. That judgment remains human.

This is where taste and discernment take over.

Enhanced Intelligence, then, is not a matter of automating everything. It is the art of composing a workflow in which delegation and friction are both used intelligently.

The Recursive Path

This is the recursive path I am now traveling.

Each new tool changes not only what I can do, but what I think is doable. Once cinematic shorts became possible, the novel itself began to expand in my imagination: first into AI-generated short films, then into chapter-based films, and eventually into a print or electronic edition where QR codes could carry readers from the page into voice, image, music, and motion.

At each stage, the affordance landscape changes.

This is why the transformation is not merely technical. It is ontological. The creative object changes, but so does the creator. The workflow changes, but so does the imagination that inhabits it.

In the old publishing ecosystem, the writer’s role was relatively narrow. The writer wrote. A publisher acquired the rights, edited the manuscript, packaged the book, and placed it into bookstores.

In today’s publishing ecosystem, the storyteller is no longer chained to that production process. With enough curiosity, discipline, and computational assistance, a single creator can compose prose, record audio, generate images, produce video, design subtitles, create music, publish across platforms, and reach audiences in multiple languages.

This does not mean everyone should do everything. Nor does it mean that traditional publishing is obsolete. It means the field has changed.

The storyteller can now become a node within an intelligent network.

That is what happened to me.

By adopting Computational Intelligence into my workflow, I moved far from the original task of writing a novel. I did not abandon writing. I discovered that writing had become one branch of a larger creative ecology.

The page was no longer the final container of the story. It had become one medium among others.

That is the transformation I did not expect. Computational Intelligence did not make me less of an author. It forced me to become one in a deeper sense.

I had to decide what to delegate and what to protect. I had to learn where speed mattered and where friction had to be embraced. I had to develop taste across media I had previously only consumed. I had to become responsible not only for sentences, but for the relations among tools, forms, platforms, and audiences.

In the end, Enhanced Intelligence was not simply a tool I used.

It was the creative field in which I learned to compose.

That is how I became the author of my workflow.