I Won’t Help Train AI

May 6, 2026

Share

I recently came across a job posting offering to pay literary writers, at the high end, $150 per hour to train AI to craft fictional stories, plays, and movie scripts. The job posting noted that “contributors will apply their literary expertise to improve AI systems’ ability to generate nuanced, high-quality narrative content. This work emphasizes strong storytelling, stylistic precision, and editorial judgment informed by published and awarded experience.” 

I’ll admit I was tempted to apply, and I even messaged the recruiter indicating I was interested. It’s not every day that a “lowly” humanities laborer is offered an hourly rate that high. And the job itself would probably be quite fun. One of the best parts of teaching creative writing is the opportunity to read others’ stories and think deeply about them on a craft level. How might we render this character more truthfully with sharper descriptions and dialogue? How might setting be used to conjure a more melancholic mood? The creative problem-solving required for such a job, even if the fiction was generated by a non-human, would no doubt be intellectually stimulating.

When the recruiter invited me to actually submit an application, I declined. I couldn’t do it. While I have philosophical issues with such a role, which I’ll get into later in this essay, on a more personal level it seemed to be an irrational act of self-sabotage. As a fiction writer, how can it possibly benefit me to train AI to do the very thing I’m trying to do? I would be selling my literary soul.

This isn’t to say that training AI in a whole host of other areas isn’t worthwhile or even beneficial. It’s just that, when it comes to the creation of narrative art, or any type of art for that matter, that’s something I strongly feel should be left to human beings. Sure, perhaps one could argue that training AI in this way allows it to better serve human creators. It makes AI a type of “fellow brainstormer,” a “companion scribe,” another “showrunner” in the writers’ room to offer ideas for character and plot. Maybe. But it’s hard to imagine if AI is able to generate narratives that can be sold to Hollywood or New York literary presses and make all those involved a lot of money in the process, that such a reality won’t come to pass. I would propose we leave AI to handle the menial and boring tasks that we’d rather not do—the routine email responses, business proposal outlines, exhaustive research reports, etc.—precisely so that we have more time to do the very things that are most human, like, well, making art.

If we’re simply gazing through the mere simulation of a mind, then we aren’t actually communing with anything.

And even if AI can craft a novel or script, perhaps even a great one, do we really want to read or watch it? I have zero interest in watching an AI-generated blockbuster or attending an AI-generated play. It’s true that I might enjoy it if I hadn’t known it was AI-generated. I might even think it’s good. But even so, I don’t care. I’d rather not spend any time engaging with it.

Art provides a lot of benefits, and one is that it allows us to commune with another human mind (or minds) in a unique way. Part of the pleasure of reading a good novel is the chance to “wear” the consciousness of the fictional characters, the narrator, and to some extent, the very mind of the one who created them. It allows us to see through another set of eyes, so to speak, and this is meaningful. But if we’re simply gazing through the mere simulation of a mind, then we aren’t actually communing with anything.

There is an interesting Black Mirror episode from many years ago that is worth considering in light of this topic. The episode, titled “Be Right Back,” features a young couple—Martha and Ash—who move to Ash’s family’s house in the countryside. Ash is killed shortly afterward. While grieving, Martha learns about a service that can compile Ash’s online social interactions and history into a realistic avatar with his voice, whom she can talk to on the phone. She opts to do this when she finds out she is pregnant, obviously feeling especially vulnerable and desiring to share the experience of having a baby with the child’s deceased father. Eventually, though, a new service is offered, whereby an extremely realistic humanoid robot can be ordered that looks exactly like Ash.

However, as with nearly all Black Mirror episodes, things inevitably go awry (though, this episode is relatively mild in comparison to others). Martha orders the robot, enjoying all of the positive qualities of her former lover. But a problem emerges: The robot has none of the real Ash’s negative qualities. It simply gives in to whatever Martha wants, essentially making the thing a sophisticated puppet. The real Ash wasn’t so amenable. The real Ash had his own desires and intentions that clashed with her own. This leads to a key moment when, realizing this simulated Ash can never replace the real Ash, Martha orders the robot to kill itself by walking off a cliff. It begins to do so, but right before it does she gets upset that, once again, it isn’t resisting her the way the real Ash would. So, what does this AI-doppelgänger do? It begs for its life, doing so only, of course, because that’s precisely what Martha wants it to do. Again, we realize, like Martha, that the robot has no real agency, desire, or will—only the simulated appearance of these things.

Story of All Stories Children's Bible
Get Your Story Bible

In a similar way, while we might be able to train AI to produce works of art that cohere to what we desire or expect—that are morally ambiguous, nuanced and compelling, reveal some complicated truth about human nature, etc.—they will, ultimately, be the product of what we have told it to give us. Just like the artificial Ash, it will merely be responding mindlessly to what we have collectively trained it to do.

But one of the great gifts of really good art is that it can be surprising. It doesn’t always cohere to our expectations or desires. It can challenge us and leave an impression on us that we could never have anticipated. And that type of art can only come from another person, from another free will and intellect capable of grasping reality in a way that no machine can. What good artists do, ultimately, is render an aspect of truth to us. And to do that, the artist must not know merely about reality, or grasp a simulation of reality, but must have personally experienced reality by virtue of living within it as a conscious being endowed with a transcendent soul. Even if we are tricked by some AI-generated work of art to believe it was created by a human, and even if we discern no subjective difference between engaging with that work of art or one made by a human being, we are still left diminished. Even if the simulated Ash was secretly swapped out with the real one, and Martha had no idea that she was now in a relationship with a machine, she is, ultimately and objectively, still not experiencing authentic communion and relationship. She is experiencing a simulation, one that is objectively lacking and dehumanized, however she might subjectively feel about it. The same is true with engaging in art generated by a machine: Even if we don’t know it or feel it, we are being dehumanized in some form, for we are not participating in the authentic type of communion that art can and does offer us.

When an artist creates something, he shares an aspect of his being. He becomes vulnerable, putting into the world a remnant of his worldview or experience so that others can share in this experience. When the resulting art is true and beautiful, it is an act of love. And we, as the receivers, are nourished by the artist’s sacrificial and loving act. We are participating in a type of communal love, benefiting from the sacrifice and service of another real human being. Sure, AI can entertain, shock, captivate, and distract us. And it’s fine for AI to do these things, I suppose, in moderation. But AI, no matter how sophisticated it gets, will never be able to offer us something that is the product of a free act of the will—something that is a sacrifice of love. And so it won’t be able to offer us what we were created for: communion with others through a shared bond of love.

That’s why I won’t help train AI to write better fiction, even for a nice hourly rate. But that probably won’t stop others from doing it, so we’ll likely have a chance to see how this more “sophisticated” AI-generated art affects our culture. We don’t seem to have the foresight to ensure that AI is ultimately promoting human flourishing. Of course, maybe AI will continue to produce hackneyed and bad writing, which it still does at the moment. But even so, I imagine that small communities of people will, thankfully, continue to push against AI-generated art, no matter how popular it becomes. Similar to those who choose to buy handmade products instead of the mass produced, I suspect that small communities will seek out the human made and authentic, nurturing our God-given impulse to create beautiful things and share them with others for the sake of love.