Process note: This is, appropriately, a piece of hybrid writing. It grew out of an extended conversation with AI in which I supplied the experiences, concerns, central ideas, and direction; asked the AI to challenge my thinking and identify perspectives I had missed; and then continued revising the argument through our back-and-forth. I am beginning to label work like this so readers—and I—can distinguish it from work I draft directly on my own.
Most of the articles I have written lately did not begin with me sitting down, placing my hands on the keyboard, and composing a clean first draft from beginning to end.
They began with me talking.
I tell the AI what I think the article might be. I explain the issue, circle the parts that bother me, contradict myself, remember something halfway through, and ask what I might be missing. I ask it to analyze the approach I am taking and show me other approaches I might not have considered. It responds. I respond to the response. Sometimes I reject what it gives me. Sometimes one small observation opens an entirely new branch. We go back and forth until the original subject has become larger, stranger, and more complicated than it was when I began.
Eventually, there is an article.
Often there are enough branches for several articles. I could separate them, and perhaps sometimes I should. But I also like the complication. I like the and in the middle of things.
An experience can be damaging and understandable. A technology can be useful and dangerous. Disclosure can be freeing and overwhelming. A person can believe something strongly and remain willing to reconsider it. Several truths can exist beside one another without one of them automatically canceling the rest.
That is where much of my writing lives. I am usually less interested in declaring that an issue is one thing than in examining how it can be this thing and that thing at the same time.
I have started calling the process that helps me do this hybrid writing.
Hybrid Writing Is Not the Same as Prompting for an Article
When people hear “AI-assisted writing,” they may imagine someone typing a subject into a box, receiving a finished article, and publishing it untouched. That certainly happens. It is one of the things people mean when they call something “AI slop.”
Slop, as I understand it, is not defined only by the presence of AI. It is defined by the absence of meaningful human attention. It is language generated to occupy space: plausible, polished, and structurally competent, but with no particular human need behind it. The person requesting it has not argued with it, tested it against experience, made difficult choices, or taken much responsibility for what it says. The work exists because producing it was easy.
Hybrid writing can resemble that work on the surface, which is where the problem becomes uncomfortable.
AI writing often goes on too long. It is eager to include every qualification, every related idea, every counterargument, and every neat conclusion. It can fit too many points into one article and then explain each of them more than once. It favors smooth transitions, balanced clauses, complete lists, and a tone that can feel suspiciously reasonable. Those have become recognizable tells.
My hybrid pieces frequently contain some of those tells.
They are often long. They branch. They qualify themselves. They try to hold several interpretations together. I could edit them more aggressively, sand down the language, break them into smaller pieces, and remove some of the completeness. Sometimes that would make them better.
But length and complexity are not foreign substances introduced into my writing by a machine. I have always returned to subjects, followed associations, connected personal experience to larger systems, and tried to see how apparently separate things belong to the same structure. My mind is recursive. I do not naturally experience ideas as isolated points. I experience them as networks.
AI is unusually capable of keeping that network visible while I think.
That can become indulgent. Complexity without hierarchy can turn into clutter. Just because a machine makes it possible to include eight worthwhile ideas does not mean all eight belong in one article. The danger is not simply that the article becomes long. The danger is that every point begins to arrive with the same weight, as though every branch were part of the trunk.
Still, I do not want to respond by pretending that a complicated subject is simple. The expansiveness of hybrid writing can illuminate connections that a tighter, more conventional argument might lose. It can help me notice assumptions in my original position, imagine how the issue appears to someone sitting elsewhere, and follow consequences farther than I might have followed them alone.
The machine increases the number of connections available. I remain responsible for deciding which connections are true, which matter, and which deserve to be shared.
That responsibility is where I locate authorship.
The Conversation Is Part of the Composition
In my process, I am not only supplying the topic. I am supplying the personal history, emotional stakes, contradictions, judgments, and recurring questions. I am also directing the inquiry.
I ask the AI to push beyond my first answer. I ask what I have failed to consider. I ask it to challenge the frame I brought into the conversation. Then I decide whether its challenge has merit. My position sometimes changes—not because a machine has delivered the correct opinion, but because the exchange has allowed me to look at my own opinion from another angle.
The thinking develops through the conversation. That makes the conversation part of the composition.
This is also why calling AI merely a “tool,” while true, can feel incomplete. A hammer does not answer back. A spellchecker does not introduce a moral complication. In hybrid writing, AI may act as organizer, interrogator, structural aid, counterargument generator, editor, mirror, and occasionally an extremely confident source of something I have to reject.
The rejection matters too. Collaboration does not require agreement. Some of the most human parts of the process occur when I recognize that the language is too smooth, the conclusion too easy, or the interpretation fundamentally wrong. The machine offers possibility. I supply judgment.
At least, that is the standard I want to hold myself to.
The Distance Helps Me Get Closer
There is another reason I like hybrid writing that has less to do with efficiency or intellectual complexity.
It gives me distance.
I tend to write about personal matters or to approach public issues through an intensely personal lens. I write about grief, mental illness, shame, family, illness, memory, responsibility, and earlier versions of myself. When I write directly from inside those subjects, the result can be very raw. That directness can be a bit much for readers. Sometimes it is a bit much for me.
The hybrid voice creates a thin membrane between the experience and its presentation. It is slightly more detached and composed. It can take something I have narrated in fragments and return it in a form I can examine without standing completely exposed inside every sentence.
That distance is not always dishonesty or dilution. Sometimes it is containment.
We already accept many artistic devices that create distance: third person, fictionalization, metaphor, structure, an editor’s intervention, or the simple passage of time. AI introduces a new and ethically complicated form of mediation, but the underlying need is recognizable. Sometimes a writer needs to step a few feet away from an experience in order to see it clearly enough to write about it.
The paradox is that AI can move the voice slightly farther away from me while allowing the article to move closer to what I mean.
There is a risk in that distance. The hybrid version may smooth away the strange sentence, abrupt confession, uncomfortable contradiction, or peculiar image that could only have come from me. AI can regulate the emotional temperature so effectively that it removes the heat the piece required.
Editing, then, does not have to mean disguising AI involvement by manufacturing roughness or sprinkling in artificial quirks. It can mean restoring the parts of my voice that the collaborative process made too polite. I can keep the structure and complexity while asking whether the finished piece retains enough of the original pressure that made me speak.
Labeling the Work
I increasingly think I have a responsibility to label these pieces more clearly.
That responsibility is partly to the audience. Readers deserve to know when the prose was substantially shaped through an AI conversation. Transparency does not settle every ethical question about training data, ownership, labor, or authorship, but it at least allows the reader to evaluate the work without being misled about how it was created.
But the label is also for me.
I want to begin distinguishing between work I draft directly and work produced through this hybrid process so I can study the difference. I want to know which qualities persist across both forms. Which metaphors are mine? Which structures appear whenever I am left alone? Does AI make me more balanced, or simply more qualified? Does it clarify my thought, or sometimes replace a sharper thought with a more comprehensive one? Which personal subjects become writable only when I have the buffer of a collaborative voice?
Over time, the labels could become a record of my developing voice rather than a disclaimer attached out of shame.
One note might say that a piece was written directly by me without AI-generated prose. Another might explain that the piece was developed through extended AI conversation: the experiences and governing ideas were mine, while AI helped test the argument, identify missing perspectives, organize the branches, and shape portions of the language.
Neither label would automatically identify the better piece. They would identify different methods.
That comparison may also reveal that some characteristics people identify as “AI voice” were part of my writing long before generative AI entered it. The recursion, repetition, connected lists, qualifications, and insistence on holding tensions together are not alien to me. My quirks and AI’s quirks happen to mesh well, which is creatively useful and analytically confusing.
If I want to understand what is most me and what is most machine, I need examples of both kinds of work.
Returning Is Not Repeating
This matters because I return to topics constantly.
I do not generally return because I have forgotten what I said. I return because I am still analyzing my opinion and direction. I want to remain open to reconsideration because circumstances change, information changes, and I change.
Our view through a window depends partly on where we are sitting. Move to another seat and something previously hidden may enter the frame. What looked central from the first position may move to the edge. The first view was not necessarily false. It may simply have been incomplete.
That is the difference between repetition and return.
Repetition says the same thing again. Return asks whether the thing still appears true. Recursion carries what was learned back through the original question and produces another version.
I think of my writing less as a collection of final declarations than as a record of developing consciousness. An earlier essay preserves what I could see from that seat. A later one shows what became visible when I moved. I do not want to erase every contradiction between them, because the movement may be more truthful than a polished illusion that I have always held my current position.
Openness does not require an absence of conviction. I can believe something strongly enough to act on it while holding it loosely enough to examine it again.
Hybrid writing gives me additional seats around the window. The exchange exposes alternative angles quickly and sometimes relentlessly. That is one reason these pieces accumulate so many branches.
But it also raises a question I can no longer answer from within the hybrid process: What does my thinking sound like now when no one immediately answers back?
Hugh Howey’s Neo Arrived at the Right Time
That is why Hugh Howey’s writing app Neo feels as though it arrived through some form of kismet.
Neo did not create my desire to try a different way of writing. I had already been thinking about assembling an outline and then writing a first draft without receiving AI-generated prose or new AI inspiration during the drafting itself. I had considered it, but I had not taken a deep dive into making it part of my practice.
Then Neo appeared: a tool for the kind of experiment I was already preparing to conduct.
I do not see this experiment as an attempt to prove that unassisted prose is purer, more moral, or automatically better. Nor am I preparing to divorce hybrid writing after defending what it has allowed me to do. Neo is a counterweight, not a replacement.
The plan is to think and outline beforehand, then enter the drafting space and write the first version myself. I want to resist asking for another angle whenever I feel uncertain. I want to leave some sentences rough, follow some instincts before they are analyzed, and discover what happens when I have to hold the structure and emotional temperature on my own.
I want to hear the voice that appears when nobody immediately responds.
The timing may be especially meaningful because I am preparing to revisit some of the most personal material I have written: cancer, my brother’s death, grief, mental health, maintenance, and earlier versions of myself preserved in earlier books. That material already contains distance because I am returning to it from another point in my life. The person writing now is not sitting in the same seat as the person who first lived or recorded those events.
Perhaps I will discover that I still need the additional buffer AI provides. Perhaps I will discover that the passage of time has created enough distance on its own. More likely, I will find different needs in different sections.
The result will give me something more valuable than an abstract position in the argument over AI. It will give me evidence from my own practice.
I can compare direct writing with hybrid writing and ask what each form makes possible. Direct drafting may restore rawness, surprise, compression, and idiosyncrasy. Hybrid writing may remain better at sustaining complex inquiry, exposing blind spots, and helping me approach emotionally difficult material without becoming trapped inside it.
Each method may teach the other.
Writing directly may help me restore more of my unmediated voice when I return to hybrid work. Hybrid work may train me to recognize structures, counterarguments, and connections when I am writing alone. The goal does not have to be choosing one permanent method. It may be learning when to use each one and becoming honest about what each contributes.
The View From More Than One Seat
I do not know where this experiment will lead. That uncertainty is part of the point.
My opinion about AI is still developing because AI is developing, the culture around it is developing, and my use of it is developing. I can value what it has made possible while remaining troubled by what it threatens. I can defend hybrid writing without pretending every use of AI is meaningful. I can recognize AI slop without accepting that everything touched by AI belongs in that category. I can enjoy the distance while watching for the places where distance becomes evasion.
And I can write directly without turning direct writing into a performance of purity.
Hybrid writing allows me to think through dialogue. Freewriting allows me to encounter thought before it has been answered, balanced, or organized by another intelligence. Neo has arrived at a moment when I am ready to put those approaches beside one another and pay attention.
The truth may not be visible from only one seat.
For now, I want to keep moving around the window.
Frank M. Anderson's Writing 
Leave a comment