The Moment Everything Got Complicated
It’s 3 AM and I’ve just finished a novel that left me feeling hollow in a way I couldn’t immediately name. The prose was competent. The plot moved forward. The character arcs resolved. Yet something fundamental was missing, and I spent twenty minutes scrolling through the acknowledgments looking for clues. That’s when I saw it: a quiet thank-you to an AI assistant “for structural guidance during the outlining phase.” I felt like I’d discovered a hairline fracture in something I thought was solid.
This is the honest truth about AI-assisted writing in 2025 and beyond: it’s not that the tools are making worse books. It’s that they’re making a specific kind of book that appeals to market forces in ways that should make readers like me nervous. And I say that as someone who has zero interest in being a Luddite about technology. I love efficiency. I love innovation. But I’m increasingly convinced we’re conflating those things with something that matters far more: the irreplaceable human struggle that makes literature feel alive.
What the Numbers Actually Tell Us
Let me lay out the landscape, because data helps us think clearly even when emotion wants to take over. According to the Authors Guild 2026 AI and Writing Survey, more than one in three professional novelists used AI tools at some stage of manuscript development last year. That’s not fringe. That’s mainstream. The tools themselves are exploding in adoption. Sudowrite, one of the major platforms designed specifically for fiction writers, saw its subscription base grow by 300 percent between early 2024 and the end of 2025. Three-quarters of those users identified as fiction writers. This isn’t a theoretical debate anymore. This is happening in bedrooms and home offices across the English-speaking world right now.
What strikes me most isn’t the adoption rate itself, but the specificity of how authors are using these tools. It’s not wholesale replacement of human writing. Nobody’s running ChatGPT-4o and hitting publish. Instead, we’re seeing sophisticated, selective deployment: structural outlining, plot problem-solving, dialogue polishing, pacing diagnostics. These are exactly the mechanical elements where AI performs most convincingly. A human being doesn’t feel like they’re cheating when they use a tool to fix a saggy middle section. And honestly? I’m not sure they should feel like they’re cheating. That’s where I get stuck.
The Disclosure Problem and What It Reveals
Here’s what troubles me more than the tools themselves: we don’t actually know how pervasive this has become. The Society of Authors AI Guidelines October 2025 now require disclosure in certain publishing contracts, but “certain” is the operative word. There’s wiggle room. A novelist could use AI for outlining, refinement, and structural analysis, never mention it to their agent, their agent never mentions it to the publisher, and by the time the book reaches us readers, there’s no trace of the process. We’re simply reading the finished thing, no receipt attached.
Amazon’s enforcement tells us something revealing too. In the second quarter of 2025, the company removed over 2,800 book listings from Kindle Direct Publishing for violating updated content authenticity policies. That’s a real number of real removal actions, which means the platform saw enough mass-market garbage generated by AI-first thinking to warrant intervention. The distinction matters: these weren’t books written by humans with AI assistance. These were AI-generated product designed to game algorithms. Yet the existence of that category has cast shadow on everything else. Readers are rightfully suspicious now. Trust is eroding.
The Expectation Problem Nobody’s Naming
Here’s what I think we’re actually wrestling with, and I need to say it plainly: we expect novels to be difficult in a specific way. We expect them to emerge from struggle, from the author’s own wrestling match with language and meaning and the unbridgeable distance between what they meant to say and what they actually managed to write. That friction creates the texture we experience as authenticity. When an author uses an AI tool to sand away problems, they’re not just solving technical issues. They’re removing evidence of their own thinking process. The book becomes frictionless in a way that mimics polish but can feel like emptiness if you know what to listen for.
Gabrielle Zevin, a novelist whose work I genuinely admire, discussed using AI tools for structural outlining on her 2025 follow-up project during an interview with The Guardian in November. She was thoughtful and transparent about it. She used the tools deliberately, not as a crutch but as a collaborator with a specific strength. And I still found myself reading her previous work differently afterward, wondering if I’d been experiencing the texture of her mind or the texture of an algorithm working overtime. That’s not her fault. That’s the trust problem AI assistance creates regardless of intent.
What Matters More Than the Technology
I don’t think the solution is prohibition or moral panic. Plenty of authors writing excellent novels are using these tools. Plenty of authors writing mediocre novels aren’t. The tools aren’t the variable that matters most. What matters is whether an author is using them to deepen their own thinking or to bypass thinking altogether. The difference is invisible in the published text, which is precisely why transparency and genuine reflection from authors themselves become so crucial right now.
What I’m asking for is honesty. Not just in acknowledgments, though that helps. But in the conversation we’re having with ourselves about what writing is and what readers deserve. Because here’s my actual position: I’ll keep reading novels that used AI assistance. I’ll keep trusting authors who use these tools thoughtfully. And I’ll keep noticing when something feels like it emerged from struggle versus something that feels like it was optimized. The former feeds me. The latter leaves me hollow at 3 AM, scrolling through acknowledgments, trying to understand what I just consumed and whether I actually wanted to consume it.
What’s your experience been? Have you noticed AI’s fingerprints on books you’ve read? And more importantly, did it change how the book felt to you, or am I overthinking this?
