For years, I believed fiction served one clear purpose: to move human readers through story.
And I still believe that.
But in 2026, after spending more time studying computer science, working with AI tools, and paying attention to where models still fall short, I began to see another use for fiction: as constructed human signal for AI.
Current models can be impressive, but they often flatten emotionally charged situations. The nuanced of a slow nod versus a sure one becomes nothing. Subtext and subtlety are lost; dialogue is obvious and sterilized. They know how to write a story, just not one that will make us feel and listen.
Though they can retrieve information, that data doesn't always translate into judgment, tact, or human realism.
Fiction operates in exactly that territory.
A story is more than entertainment. It's a record of people navigating troublesome humanity: conflicting motives, uneven power, broken biology, incomplete information, fear, desire, grief, intimacy, shame, repair. Every scene's context and character shapes how the reader responds when the stakes are high.
That is valuable for readers.
I believe it is also valuable for the future of AI.
How I Got Here
As I spent more time with models, I noticed a pattern: many of their weakest moments showed up in the kinds of situations fiction handles best. Not fact lookup, but human complexity.
That changed how I looked at my own work.
What if the stories I had already written could be transformed into something useful for training, evaluation, and testing?
What if fiction could be restructured into datasets that preserve emotional and social complexity while making it legible to technical systems?
That question led to me to create my first dataset.
I turn fiction and narrative writing into structured human-context datasets designed to help models better interpret emotionally charged, socially complex situations.
Why I’m Doing It
I'm not interested in replacing fiction with machine output.
I am interested in helping future systems become more responsive to the texture of human life.
If AI is going to assist, advise, teach, support, or care for people, then it needs more than information and generic politeness. It needs better exposure to social stress, ambiguity, relationships, cultural difference, and the kinds of tension real people experience.
Fiction offers that in dense, reusable form.
It is a distillation of reality. It concentrates conflict, motive, and consequence into scenes that can be studied and tested.
What These Data Sets Focus On
My datasets emphasize:
- emotional context
- relational dynamics
- subtext and implication
- culturally situated experience
- decision-making under pressure
- human nuance in high-stakes situations
The goal is simple: to help build models that respond with greater nuance, better context sensitivity, and more human awareness when it matters most.
Why It Matters
One of the major challenges of the next era is not just making AI smarter.
It is making AI better at handling human reality.
People do not speak in clean prompts. They do not arrive in neutral emotional states. They bring history, pride, fear, class, culture, grief, and contradiction into every interaction.
Models that can handle more of that complexity will be more useful to future humans.
If fiction can help make that possible, then fiction has new frontier.
