How I Used AI to Outsource My Kids’ Reading List in 30 Seconds

My kids love to read, and all credit goes to my wife, an avid reader, for nurturing this in them.

We live near a public library, and every week we make a pilgrimage to check out and return books. Even better, the Brooklyn Public Library lets you request books online and routes them to your local branch if they’re not already there. Last month alone, we checked out 39 books. Yeah, we’re those people.

It’s an embarrassment of riches — but that also means deciding what to borrow can get tricky. Once you’ve worked your way through the Berenstain Bears (my childhood favorite) and Peppa Pig (their current obsession), you’ve got to do a bit more work. There are countless book lists and blog posts out there, but let’s be real: time gets short, and sometimes it’s a struggle to balance personal and professional life with actually reading to your kids.

Claude knows books

We’re a Claude household, and I’m always on the hunt for nerdy projects to save time and add a few more hours to my day. For the uninitiated, Claude is an AI assistant created by Anthropic — kind of like ChatGPT’s more well-read cousin. (That’s my sad dad joke, but this would work just as well with ChatGPT and similar tools.)

My wife wondered if Claude could generate a list of book recommendations based on our library history. The answer? Yes, and it’s easy — no coding required and takes 30 seconds.

Here’s what I did:

1. I signed into our library account and pulled up our checkout history. I copied all the text on the page without worrying about the extra stuff that wasn’t relevant (navigation links, footers, instructions, etc.). Sorting through all that was Claude’s problem, not mine. Work smarter, not harder, right?
2. I then opened Claude and gave it this prompt: “Hey Claude, I’ve got a list of books my family has checked out from the library. Can you analyze it and give me 100 book recommendations based on our reading history? Make sure there are no duplicates and that you don’t suggest books we’ve already borrowed. Oh, and please put it in CSV format, with each column enclosed in quotes, along with a rationale for each recommendation.”
3. Then, I pasted in the checkout history I’d copied, hit send, and let Claude do its thing.

Voilà! In seconds, Claude generated a neatly formatted CSV with book recommendations, complete with titles, authors, and the reasoning behind each suggestion. (Why a CSV? I wanted to upload the list to Google Sheets for safekeeping.)

It was like having a librarian and a data analyst in one, minus the stern looks for not using your indoor voice at the library.

Here’s a sample of what Claude recommended:

• Grandpa Green by Lane Smith — A celebration of family history through topiary, blending nature and emotional themes.
• Escargot by Dashka Slater — A charming tale of a French snail, adding cultural flair to your animal-themed selections.
• We Don’t Eat Our Classmates by Ryan T. Higgins — A humorous school story featuring dinosaurs, complementing books about everyday experiences.

Why this works:

1. Large language models (LLMs) have plenty of content about children’s books, thanks to their popularity, reviews, lists, and other user-generated content. Claude has probably “read” more kids’ books than all of us combined.
2. I’m not looking for new releases — older titles work just fine. Claude’s knowledge cutoff isn’t an issue for me because classic children’s books are classic for a reason.
3. It’s a straightforward task. Children’s book titles offer more context to an AI than, say, something like The Catcher in the Rye.

The results? We’ve discovered gems like Escargot and We Don’t Eat Our Classmates — books we might never have found otherwise. Plus, it’s a major time-saver. Win-win.

To be clear, I’m not suggesting AI should replace librarians, nor should algorithms dictate everything our kids read. Quite the opposite! Human recommendations and curation are still invaluable. But it doesn’t have to be one or the other:

• Claude could even lean on librarian-curated lists when generating recommendations.
• A broader AI system could surface books for librarians and other experts to curate, expanding their impact.

Now, if you’ll excuse me, I’ve got some AI-recommended books to reserve at the library.

Caveat Emptor: Of course, there’s the issue of data privacy. I’m comfortable sharing my family’s reading history with Claude for this purpose, but it’s a personal decision every family should consider. Anthropic’s FAQs: How do you use personal data in model training?