I’m a children’s librarian, and by the nature of the trade, there is nothing I love more than a theme. In fact, if a theme is not provided, sometimes I make up my own for the thrill of it. I do not simply pick good books for storytime. I have an uncontrollable urge to pick books about bugs, unicorns, or aliens. When my library turned fifty and I was faced with a lack of “anniversary of a public institution” books, I decided it was a “library birthday party” instead—and themed everything around birthdays!
So it should come as no surprise that I am the biggest fan ever of the Collaborative Summer Library Program (CSLP). Hundreds of libraries across the country using the same theme? Where do I sign? Another great advantage of CSLP is the resources it provides: activities, decoration ideas, and, best of all, books that fit the theme.
That is why I was so confused last year when I earnestly thought I had ordered one of its recommended books, only to be met with one of the most carelessly conceived picture books I have ever seen. The rhymes were half-rhymes, any semblance of a plot escaped me, and the art was obviously generated using AI.
I looked back at CSLP’s recommended reading list and realized my mistake. I had meant to order The Color Monsters by Anna Llenas (a genuinely charming picture book) and instead mistakenly ordered The Feelings Monsters. I refuse to name the “author” of this book because they have “written” and published more than sixty picture books in under three years. I was not alone in my mistake. The Feelings Monsters is listed on Amazon as a “best seller” when one types The Color Monsters into the search bar.
Unfortunately, with the growing ease of “making” a picture book using AI, I suspect the future will be full of these intentionally confusing titles as the AI slop-jocks chase one predictable trend after another. A quick search on Amazon shows that the current “gold rushes” for shoddy picture books are the World Cup, America turning two hundred and fifty, and dinosaurs—the CSLP theme for this year.
The issue isn’t self-publishing, KDP, or the fact that people without expertise are attempting to make picture books. The issue is that I have yet to read an AI-generated picture book that I feel is worthy of any child’s attention. I have never opened one of these books and felt the urge to push it into the hands of a child. Instead, I am overcome with the desire to run it through the nearest shredding machine.
AI operates according to the law of averages. When you ask it to “write me a book about a bedtime routine featuring a bunny,” it scours the internet for similar ideas and titles (while also drawing indiscriminately from blogs, Reddit threads, and advice columns) and produces one glorious nothing-burger of a book.
There are several problems with the growing number of these sanded-smooth, seemingly flaw-free plots and eye-wateringly shiny “books.” The first is that the child is the last consideration in both the prompts and the books themselves.
We have come a long way from Fred Rogers’ philosophy about how to speak to a child. Now, no human being is even thinking carefully about the words we are about to read to children. The terrible book I accidentally ordered not only failed to rhyme correctly, but the language model also seemed fixated on the word “hate.” How is that an appropriate word to use repeatedly while trying to teach children about their feelings?
Furthermore, great picture books tend to solve a problem sideways, with a sleight of hand and nuance that AI is not capable of producing. While it might be “intelligent,” it is not capable of thinking like a person. Goodnight Moon is not a series of self-soothing rituals stolen from a parenting blog. Guess How Much I Love You is not a listicle about attachment styles with actionable steps. Where the Wild Things Are isn’t an instruction manual for emotional regulation.
The bumbling robot may be able to scrape the web for information, but it does not know what a child needs to hear. It does not know what will capture a child’s attention or affirm their lived experience. It will forever lack the human touch needed to create true art.
We are witnessing the death of expertise. Twenty years ago, someone who wanted to write a picture book was probably continuously exposed to children’s literature, even in adulthood. They may have been a teacher who adored the time they spent reading to children. Perhaps they were a devoted parent who spent years curating the perfect library for their family. Or perhaps they were simply an ordinary person who felt passionate about young readers and knew how magical a good book could be.
These authors had to sit down and troubleshoot what did and did not work in a good picture book. They could draw inspiration from their favorite books and abstract what they had learned in a way only the human mind can. They almost certainly asked other people for their opinions on the manuscript before publication. Through this effort, they produced something worthy of a child’s time and attention.
A person can write an obscure allegory about the human condition and whittle it down into a picture book appropriate for children. A language model can only copy that idea, tweak the story, attempt to make something “new,” and miss the entire point.
To put it simply, AI will never get “it.”
“It” is the feeling you get when you read to a child and realize they are completely captivated. “It” is the demand for the story to be read again. The “it” factor is what makes you pleased that you chose this particular book to read, even when you cannot completely quantify why.
No good picture book is born from the law of averages. The best books in the world are delightfully weird in ways no one could have predicted. Who would have thought that a dark noir story about terrifying carrots would become a bestseller? Or that the story of crayons attempting to unionize would become a perennial favorite?
These are admittedly bizarre ideas, but the care and thought put into them transformed them into delightful stories I am always happy to read, and my audience is always happy to hear. A computer algorithm can never replace the time and care someone puts into transforming a strange idea into a fabulous story.
Artificial intelligence’s inability to be wholly original begins to create an ouroboros that should concern anyone who cares about children’s literature. The first wave of AI-generated picture books at least attempted to imitate books written by people. As these books accumulate online, however, the machine is increasingly surrounded by its own output. It is no longer merely borrowing from Giraffes Can’t Dance or The Very Hungry Caterpillar. It is borrowing from countless machine-written stories that were themselves assembled from other books.
Imagine making a photocopy of a beautiful painting. The copy is recognizable, but a little duller than the original. Now photocopy the photocopy. Then photocopy that one. Eventually, the image is still there, but the fine details that made it beautiful have disappeared into gray fuzz.
Children’s literature risks the same fate. A derivative of a derivative, containing no unique thought or craftsmanship, is a worthless object, unworthy of print or anyone’s attention.
Ironically, the people trying to use AI to crack the code of bestselling picture books seem to be discovering that this approach does not work very well. While copying books that appeared on the CSLP reading list last year produced a few “Amazon bestsellers,” the same cannot be said this year of the dinosaur books that were hastily slopped out.
Rather than abandoning the experiment, these “authors” who are willing to do anything but write have simply shifted their research onto real human beings. Moderators of children’s literature communities have begun noticing an influx of vague questions that read less like genuine curiosity and more like market research simple enough for an infantile robot.
“What makes a good picture book?”
“What themes do children enjoy?”
“What books does your toddler ask for again and again?”
These are wonderful questions when asked by an aspiring author hoping to learn the craft. They become rather unsettling when the answers are turned into nothing more than another prompt to feed to the machine.
The assumption behind these questions is understandable. If enough librarians, parents, and teachers explain what makes a picture book successful, surely a language model can combine those answers into the next bestseller.
Except that is almost the opposite of how memorable books are born.
If you asked one hundred librarians to list the qualities of an excellent picture book, you would likely receive thoughtful answers about unique premises, rhythm, emotional resonance, strong illustrations, and child-centered storytelling. Those are all true. Yet none of those answers would have predicted Creepy Carrots!. None would have predicted a pigeon desperately wanting to drive a bus. None would have suggested a tale about a frantic family trying to bathe a cat.
The ingredients are not the recipe.
This is the peculiar thing about expertise: it can often recognize excellence far more easily than it can manufacture it. Ask a master violinist what makes a beautiful performance, and they may speak eloquently for an hour. That does not mean someone can type their lecture into a chatbot and emerge as a concert violinist.
Children’s literature is much the same. There is no secret checklist hiding in a Reddit thread. There is no perfect prompt buried in a Facebook parenting group. There is only the slow work of reading deeply, observing children honestly, revising relentlessly, and making hundreds of creative decisions that no algorithm can reduce to “things children like.”
Perhaps that is why the books that endure are almost always a little strange. They do not feel optimized. They feel personal. Somewhere, someone had an idea that seemed delightfully odd and loved it enough to spend years making it as good as it could possibly be.
Children deserve those books. Not only because they are written by humans as a matter of principle, but because someone cared enough to think beyond the average. The average child has never existed. Every child who sits cross-legged on the storytime rug is wonderfully, gloriously specific.
They deserve books made with that
Want me to critique your picture book? Become a founding member.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.