The Numbers Everyone’s Talking About (But Missing the Point Of)
Let me start with what’s actually happening in the literary marketplace right now, because the numbers are genuinely staggering in a way that deserves real conversation instead of hot takes. According to the Alliance of Independent Authors AI publishing report, Amazon KDP saw approximately 700,000 AI-assisted or AI-generated books uploaded in 2024. That’s not a typo. For perspective, that’s a 300 percent increase from 2023. The sheer volume is almost impossible to wrap your head around until you think about what it means: if you’re browsing KDP romance or thrillers right now, there’s a very real chance you’re swimming through an ocean where more than three-quarters of new releases were generated by algorithms rather than written by human hands working through human problems.
Amazon implemented new content disclosure requirements in September 2024, technically requiring authors to declare AI-generated material. But here’s where the story gets less reassuring. The enforcement is spotty at best. Authors Guild critics argue—and I think they’re right—that self-reporting with minimal verification creates a system that’s more checkbox than checkpoint. The people uploading these books don’t always disclose. The platform doesn’t always catch them. And readers? We’re mostly clicking blindly.
What This Means for Writers (And It’s Not Pretty)
I want to be fair here. The Authors Guild 2025 income survey polled 5,000 writers and found that 62 percent reported lower income compared to 2022. That’s more than three in five professional authors with declining earnings, and the primary culprit cited by respondents wasn’t changing reader habits or economic shifts. It was market flooding with algorithmically generated fiction that costs nothing to produce, requires no revision cycles, and generates just enough engagement to push human-written books further down search results. A writer friend texted me last month that her thriller, which took eighteen months and required her to quit a part-time job to finish, climbed to position 847 in its category on release day. An AI-generated competitor with similar keywords and a higher upload volume (because, again, free to make) sat at position 34.
This matters beyond economics. We’re watching the professionalization of writing become harder to justify financially, which means we’re also watching the demographic of who gets to be a novelist slowly shift toward those who can afford to write for free. That has implications for whose stories get told, which is never just a literary concern.
But Here’s What’s Genuinely Interesting: We Can Actually Feel the Difference
This is where my midnight-text energy comes in, because this part genuinely fascinates me. A January 2025 study from the University of Edinburgh published in the journal Cognition used fMRI imaging to measure reader brain activity while people read both human-authored and AI-generated fiction. The researchers found measurable, quantifiable differences in how readers’ brains responded to the two types of texts. Specifically, they measured empathy activation and something called “narrative transportation”—that state where you’re so deep in a story that the outside world disappears. Both showed lower activation when readers encountered AI-generated fiction, even when the plots, pacing, and prose quality were comparable on the surface.
Let me be clear what this does and doesn’t mean. It doesn’t mean AI fiction is objectively “bad.” It means something more subtle and more interesting: our brains register something different at a neurological level. We’re not imagining the uncanny valley feeling. We’re measuring it. And the question that keeps me awake at night is this: what exactly is our brain detecting that algorithms haven’t figured out how to replicate? Is it the fingerprint of human struggle embedded in narrative choices? The scar tissue of real experience? Something about the way actual writers have to sit with uncertainty for months while an algorithm just samples from existing patterns?
Following the Genre Breadcrumbs (And What They Tell Us)
Edifix’s 2025 analysis found that romance and thriller genres account for an estimated 78 percent of identified AI-generated KDP uploads. Which makes sense, right? These are genres with established formulas. They have recognizable beats. They’re the most mathematically predictable corners of fiction. And here’s the thing that haunts me: those genres also contain some of the most emotionally sophisticated, psychologically complex contemporary literature. The best modern thriller writers are doing profound work with suspense as a vehicle for exploring trauma, paranoia, the fragility of truth. The best romance writers are dissecting intimacy, vulnerability, power dynamics, the terrifying act of choosing another human. These aren’t simple genres. But they’re the ones most vulnerable to algorithmic flooding because they’re also genres where readers engage with patterns, and patterns are exactly what AI is built to reproduce.
What Happens Next (And What We Get to Decide)
I don’t think the answer is nostalgic hand-wringing or pretending that pre-internet publishing was some golden age of literary purity. It wasn’t. There were always books written for profit rather than passion. There were always gatekeepers and arbiters and whole categories of writers who never had access to the system. What’s different now is the scale and the speed. A human being can only write so many forgettable books in a lifetime. An algorithm can generate them endlessly, instantly, at zero cost.
But here’s what I keep coming back to: readers know. Not always consciously. Maybe not always in language they can articulate. But we know. Our brains know. And that knowledge creates this strange moment where the line between human-made and algorithm-made art isn’t about snobbery or gatekeeping anymore. It’s becoming measurable, documentable, neurological. We’re not just preferring human fiction because of some romantic attachment to authenticity. We’re responding to something real that human consciousness creates when it works through story.
So where does that leave us in 2025? Probably in uncertain territory for a while. More disclosure requirements, hopefully. Better filtering tools, possibly. Readers becoming more intentional about seeking out verified human authors, maybe. But also this: a real conversation about what gets lost when we algorithmically generate fiction to fill every available market niche. It’s not about gatekeeping or snobbery. It’s about preservation. It’s about whether we want to keep a world where ordinary people can spend years agonizing over sentences and have that labor mean something economically. It’s about whether narrative itself, this ancient and irreplaceable thing humans do, deserves protection from being turned into just another content commodity.
I’m genuinely curious what you’re seeing in your own reading life right now. Are you running into obvious AI fiction? Does it feel different to you? Or does it all kind of blur together in the endless scrolling? I’d love to talk about it, the kind of conversation that matters, not the kind that has winners.