(Don’t Create) the Library Torment Nexus

I have been doing some deep thinking about the use of AI in Libraries, and like many other colleagues I reached the inescapable conclusion that there is currently no ethical way to use generative AI at all, let alone within Libraries.

As information professionals, Library Workers have historically and contemporaneously firmly positioned ourselves on the side of copyright protection and against piracy and intellectual property theft. It is widely documented that generative AI programs have been trained on illegally downloaded and stolen materials so I am constantly being surprised and horrified when I see colleagues gushing about how much fun they are having learning to use generative AI in their libraries, “It saves so much time”, “I no longer have to think too much about creating lesson plans”. I have been urged not to worry too much about it as AI is the wave of the future and that there is a push to get governments to regulate the companies creating AI programs to make sure that AI training is done legally, but the genie is out of the bottle on that one and barely a day goes by without authors and artists posting more evidence that their works have been stolen by multi-billion dollar organizations to feed the beasts that are their generative AI programs.

Libraries are not immune to the looming threat of generative AI, there have already been reports of AI slop (this is a real term for works created by AI trained on pirated works) appearing on digital platforms such as Hoopla that are available through libraries. I fear that in the not too distant future library services will incorporate AI programs to complement and eventually cut down on paid staff.

Databases containing information about read-alikes, pen names, ghost writers, genres as well as every user-generated tag and category you can think of will be fed into a program that will be able to spit out recommendations when a patron types in a question looking for more books in specific genres or on certain subjects. If such a program is integrated into library management software it will eventually be able to create a patron profile and generate suggestions based on a patron’s reading history and general preferences as well as their use of library digital media and what they access online through library computers. The potential for privacy violations aside, this type of program may end up funneling readers into a reading cul-de-sac, keeping them reading books that they like, which may not seem like a problem but this could prevent them from finding new authors, or reading outside their comfort zones and broadening their reading horizons.

If library services are outsourced and run by for profit entities that own publishing companies, it is entirely possible that an AI program will eventually have enough data scraped from library users to determine the most popular items loaned out by the library; books and stories can then be algorithmically generated, then published by the publishing arm of the company (or just be made available in e-format to keep physical costs down), the local authority that pays the company to run their libraries will be charged for these slop titles that are placed into the collection. They will be catalogued by another AI program and made generally available for library users, eventually it will become a perfect circle of AI generated works based on other slop works that were once based off actual works created by humans.

This horror story may seem like the stuff of paranoid fantasy, but with the current push for AI in all industries it is only a matter time before it is integrated into pubic services and services that once relied on human interaction will be rolled into an customized library AI program.

Comments are closed.

Post Navigation