Image Source: Nana a Mischevious Monkey by Shum Zay on Amazon
Blog Post #12: AI Books and Libraries
By Madeline Feehan
Is an AI author really an author? Are AI books valid, and should they be on library shelves, or should we steer clear of them? What about AI-assisted work? Is the best compromise adding some to our collection and clearly labeling them as AI, or is it best to not even indulge at all? Do you find work that was written with the help of AI but was also written by a human less egregious than something entirely written by AI? Should those books still be labeled? These are some of the many fascinating questions discussed by guest speakers who came to my library. Lisa Zuena, Nicole Scherer, and a third librarian from the Nassau County Library Association came in to talk about AI slop and what it means for the future. AI slop refers to poorly edited, weirdly formatted, strangely written books that find their way into ordinary collections. This happens when librarians order books without doing the proper fact-checking, and the workshop focused on how we can steer clear of that. The workshop also referenced bootleg DVDs (often full seasons of brand new TV shows like Only Murders in the Building and Rings of Power that have yet to be released as box sets) and other forms of digital media that libraries should keep out of their collection for copyright reasons.
This connects to social media, because we have many free tools at our disposal that can help us a) identify if a book/writing is AI, b) judge the quality of AI books, and c) scan our catalogs for AI content. Platforms like Quillbot and GPTZero allow you to put 100 words in, and they analyze how much of the text is written by AI. There are also patterns you can look out for. If a book is published independently, that doesn't mean it's AI....but it's still something to take note of because most AI books are self-published. Other telltale signs are generic language, particularly in the author blurb and dedications. For example, an AI-generated book shown to us in the panel had a dedication that was as generic as possible: "I'd like to thank my family, my mentors, industry experts," without any specification. You can also reverse image search the author to see if it's a stock photo or a real person. You can type in a person's name to see if they have any kind of digital footprint, previous publications, website, news articles, or any other available information. Another thing to look out for is the frequency of the author's publications. For example, the librarians showed us an AI author on Amazon who published a new book every month, sometimes multiple in the same week, which was a huge red flag (we weren't surprised when his author picture turned out to be a stock photo). AI books are also typically formatted oddly and don't look like your 'typical' book, whether the font is clunky, the index looks weird, or there are typos and logical flaws. Just to show you how poorly made these books can be, the presenters showed us an example of a finance book that randomly had five chapters in the beginning about puberty, and my personal favorite: a book about traveling to Melbourne that has 20+ pages of vampire-themed coloring pages at the end for no explicable reason.
My other personal favorite example was AI-generated picture books. The 'prolific' Amazon author Shum Zay wrote two beauties: Nana a Mischevious Monkey and Lulu Goes to Trip (yes, that is the actual title, and the inside has just as many wonderful misspellings). Between odd, mismatched clipart, stories that don't make any sense, and egregious spelling and grammar issues, these books are anything but quality. However, they serve as great examples of shame and can teach us exactly what to look out for. The first red flag should be the author's bio: "I'AM SHUMZAY, A STORYTELLER AT HEART AND A CHILDRENS AUTHOR BY PASSION.INSPIRED BY MY LOVE FOR FAMILY {MY HUSBAND AND KIDS} FRIENDS AND COMMUNITY, I WRITE STORIES THAT CELEBRATE THE IMAGINATION AND CHILDHOOD.READING OR WRITING BOOKS IS MY FAVORITE JOB."
Source: Shum Zay's bio on Amazon
This bio is a walking red flag, with the randomly all-capped last name, the improper grammar, and the cherry on top: the {my husband and kids}, which seems like a prompt in progress that the person using the AI forgot to fill in the blanks of. The other red flags are the lack of a publisher, the book's strangely worded descriptions, and the surprisingly cheap price of the books. Not to say a legitimate book couldn't be found for $8, but that, combined with the barely legible sentences, should give you a clue: "THIS BOOK IS ABOUT A TINY STAR WHO IS INTERESTING IN EXPLORING THE EARTH, AND FINALY GOES DOWN TO EARTH, AND SHARES WITH US WHAT HE SAW AND EXPERIENCE THERE.BOOK IS FILLED WITH COLORFUL ILLUSTRARTIONS ENHANCING THE BEAUTY OF STORY."
Source: Lulu Goes to Trip by Shum Zay on Amazon
How should librarians and libraries respond to the AI author phenomenon? The big publishers we order books from don't always have safeguards and transparency, so AI books can slip through the cracks. The one thing we should do, regardless of where you stand, is train yourself to spot AI books. This will help you avoid purchasing them, and can allow you to help patrons that might fall for its trap- even if they don't care and still want to take out the book, AI or not, it's good to let them know it's AI. We can raise awareness of AI books on social media and post our library's stance, or we can use social media behind the scenes to evaluate books and detect if things are AI for ourselves. You should also use the Internet to see what the public, and in particular, your community's reaction to AI is. If it's favorable, consider an AI shelf with labeled books. If it's negative, keep weeding through and make sure AI books don't slip into orders.
I hope you enjoy the cursed examples I've shared, and I hope this post has made you aware of AI red flags.
Comments
Post a Comment