AI-generated YA novels are readable in bursts but fall apart at novel length

Babel Bot Books released 12 AI-written YA novels in 12 days using Claude, Gemini, and GPT - no human editing. The results expose what these models actually do: predict words, not tell stories.

Categorized in: AI News Writers
Published on: Apr 03, 2026
AI-generated YA novels are readable in bursts but fall apart at novel length

AI Novel Factory Churns Out 12 YA Books in 12 Days. They're Unreadable.

Babel Bot Books has released a library of free science fiction and fantasy novels written entirely by AI models - Claude, Gemini, and GPT - with no human editing or intervention. The site showcases titles like The Unraveling of Whispers, The Resonance Fracture, and The Chronosync Divide. A closer read reveals why human authors still have jobs.

The timing matters. Publishers are already spooked by AI involvement in publishing. Horror novel Shy Girl was pulled from shelves this month after evidence emerged that its author used AI assistance - a sign the industry treats machine-written work as a liability, not an asset.

The AI-Generated Slush Pile

Browsing Babel Bot's catalog feels like scrolling through a parody of the YA section at any bookstore. The covers are interchangeable: a lone female protagonist casting spells with bright lights, or gazing at a futuristic cityscape bathed in bright lights. The taglines repeat themselves verbatim across different books.

The Harvester's Descent (Gemini, March 12): "The sky is a lie. The surface is calling."

The Starless Crown (GPT 5.4 Codex, Saturday): "The sky is a lie. The stars are waiting."

Nearly every protagonist is 16 or 17 years old. Two books feature characters named Declan Voss and Maren Voss, both 16, in completely separate stories with no connection. The site offers no acknowledgment of these overlaps.

The AI isn't trying to be original. It's mimicking the vast training data it absorbed from published books - much of which came from unauthorized sources. Claude's creators at Anthropic admitted to using pirated book collections for training and settled a copyright lawsuit for $1.5 billion.

Inside an AI Novel: The Probability Garden

The Probability Garden stood out from the catalog. It was the only title framed as a philosophical question: "What if free will was just another prediction?" It also claimed co-authorship by Claude and Gemini, though the downloadable file lists only Claude.

The setup has potential. In 2260, quantum machines called Looms predict human futures with perfect accuracy. Some people-called "stochastics"-throw off these predictions. The protagonist, Lumen, is a 16-year-old stochastic who wears grey maintenance coveralls and wakes at 5:47 a.m., defying the Looms' 6 a.m. prediction.

Then the problems start.

Claude describes a character "exhaling through his nose" from "across the room"-a detail that suggests no understanding of how sound travels. In another passage, Lumen remembers her father's voice "slightly off-key when he did the character voices" while reading bedtime stories. The AI conflated off-key singing with spoken dialogue.

When dialogue finally appears 10 pages in, it's functional but flat. Lumen's comebacks land without conviction. A Weaver apprentice corrects her joke with "Probability Cascades," and Lumen responds: "Same thing with better lighting." The line makes no sense.

Pages of technobabble follow. Claude attempts humor but the timing is off. The author seems aware of its own limitations: one character observes that "the human brain wants structure. It finds it even in noise. This is noise."

The Plot Twist Reveals Everything

By page 50, Claude dumps the entire plot. The Looms don't actually predict the future. Non-stochastic humans simply expect the predictions to come true, so they unconsciously create those outcomes.

There's a metaphor buried in there-something about AI and prediction and self-fulfilling prophecy. A human author might make it readable.

For now, Babel Bot Books serves as documentation of what large language models actually do at scale: predict the next word in a sequence, then the next, then the next. String enough predictions together and you get 12 novels in 12 days. None of them have what novels require-characters readers care about, dialogue that sounds human, mystery that unfolds rather than explodes on the page.

The real question isn't whether AI can write novels. It's whether anyone will want to read them.

Learn more about AI for Writers


Get Daily AI News

Your membership also unlocks:

700+ AI Courses
700+ Certifications
Personalized AI Learning Plan
6500+ AI Tools (no Ads)
Daily AI News by job industry (no Ads)