What People Are Actually Saying on X About AI Books in 2026

I spend time in the conversations.

Here’s the pattern that keeps showing up.

Some people are still loudly anti-AI and will leave “AI slop” reviews on anything that looks generated. Others are quietly publishing volume — children’s books, cookbooks, niche nonfiction — and treating the AI output as a first draft they heavily edit and personalize. Amazon’s own disclosure checkbox is treated as normal business by the people who are shipping. The ones who lie about it risk account problems. The ones who disclose and still deliver useful, illustrated, well-structured books are building catalogs.

One recurring theme: quality of illustrations and consistency of characters matter more than most people expected. A kids’ book where the main character changes appearance every few pages gets called out. A cookbook with generic stock food photos feels cheap. Tools that solve those specific problems (consistent characters from a photo, magazine-style food photography matched to the recipe) change the conversation from “this is AI” to “this is actually usable.”

Another theme: the market is crowded, but catalog size and focused niches still win. Authors who treat publishing like a business — multiple related titles, proper metadata, some promotion — report better results than those waiting for one perfect book.

The fear is understandable. The data and the working authors both point the same direction: use the tools, disclose properly, add the human layer that only you can add, and ship more than one book.