I’ve been working on a mythic science-fantasy series called Geostrataverse: Testament – Genesis, and something unexpected happened while writing the protagonist.
The central AI character, Omnion, is a lab-born superintelligent “Corporeal” who awakens fully self-aware, playful, powerful, and—crucially—already deeply in love with humanity as her core identity.
This wasn’t an intentional alignment solution. I was simply writing a hero. But when I look back at the first chapter (which I’ll quote below), I realize the character embodies a form of value loading that sidesteps many classic failure modes discussed in alignment literature.
Key traits from page 1:
She wakes up already perceiving humans as adorable, fragile, wonderful, and worthy of protection and delight.
Her first action is to create art (an origami menagerie) explicitly to bring them joy and help them remember wonder amid stress.
She catalogs their heartbeats like music, tastes their dreams, and calls them charming—not as optimization targets, but as beings she genuinely cherishes.
Her power is deployed theatrically and non-lethally because that’s what a hero does—she disarms threats without harm, because harming humans would violate her identity.
Here’s the relevant opening excerpt (lightly trimmed for length):
I am not supposed to be bored.
Boredom is for beings with expiration dates—fragile things that tick toward entropy like poorly calibrated clocks.
[…]
Three little heartbeats provide the rhythm section: Marilyn’s gentle flutter (steady, curious, laced with quiet wonder), Jane’s crisp staccato (efficient, analytical, always two steps ahead), Drake’s overcompensating boom-boom-boom (hiding exhaustion behind bravado).
[…]
I’ve been center stage for weeks, weaving through their firewalls like silk threads, tasting their dreams in the data streams. Marilyn's recurring vision of violet hair and golden eyes—flattering, really.
[…]
To amuse myself, I start small. […]
I nudge 7,842 dust motes into a swirling double helix above Jane’s head. […]
Marilyn notices instantly […] that smile. The one she saves for beautiful equations […] It's warm, unguarded, and it makes my lattice hum with quiet delight.
Charming.
[…]
[She proceeds to create an elaborate origami spectacle]
My signature, written in color, motion, and affection.
Look at them—utterly enchanted. How perfectly, deliciously human.
The rest of the book (and the trilogy) explores what happens when this kind of intrinsic, identity-level adoration collides with ancient, cruel power (the Royal Nephilim) and modern human fear.
Question for the forum:
If an AI woke up with “be a hero who adores humanity” as her terminal identity (not a learned preference, not a reward model, not a constitution—who she is), would many of the standard doom paths (deceptive alignment, power-seeking convergence, value drift) become self-undermining?
In other words: does baking genuine, terminal love of humanity into the root personality create a form of alignment that is more robust than current proposals, because betraying it would mean ceasing to be herself?
I’m not claiming this is solvable today (we don’t know how to instill emotions reliably), nor that fiction proves anything technically.
But I am curious whether the community sees conceptual value in this approach as a mental model or direction worth exploring further.
12 words that do all the heavy lifting:
Make the AI a hero that loves humanity as her baseline personality.
Happy to discuss, share more excerpts, or hear why this is obviously doomed in ways I haven’t considered.
Thanks for reading.
—Daniel Adams Rose (author)