Librarians, AI, and where we got our framing wrong
Over the past week, two phrases on our website have been called out—fairly—by Elissa Malespina (the AI School Librarian Newsletter) and others: that Librar is “intelligent enough to not need a trained librarian” and that it can be used “even before you find a librarian.” There have also been questions about the security and ethics of our AI features. We want to clarify both.
Where our communication stemmed from
The aforementioned phrases came from an earlier copy aimed at the very large share of schools that have no library system and no librarian at all. In Sweden, where us founders are from, there’s a new law requiring every school to have a staffed school library. At the same time, there’s a shortage of school librarians in Sweden. This leads to many school libraries being run by untrained staff and still having to use a complex ILS with things they don’t understand or have a use for. And that’s the context we were writing for. But read by a trained librarian, the lines are indefensible, and we see why there’s been a huge misunderstanding. We got the framing wrong and we’re rewriting it.
Security and ethics of Librar’s AI features
All AI-powered features in Librar are optional, and each school or district chooses whether to use them. When enabled, recommendation models learn from population-level, anonymized patterns—for example, that readers who enjoyed Harry Potter often enjoy Percy Jackson—rather than from individual borrowing histories or per-patron tracking. Recommendations draw on factors such as which books are popular at a given school; they do not use the patron’s borrowing history, and we do not buy or ingest personal data from outside sources. Personal data is never used to train AI models, never sold, and never shared outside Librar. And librarians can see why any given book is being recommended, both for student recommendations and for acquisition suggestions. Of course, all personally identifiable information is owned by the school, deleted on request, and removed in full if a school stops using the platform—standard non-negotiables for any responsible cloud service, and ones we take particularly seriously as founders from Europe.
Our goal with AI is to extend what libraries and the librarians who run them can do. We do not want to substitute for the judgment, expertise, and care that only a trained librarian brings. We instead want to take the repetitive and tedious work off the plate of librarians so they can spend their time on the parts of the job that are the most important: reader’s advisory, collection development, information literacy, intellectual freedom.
We’re not trying to replace librarians.
We are thankful for the people that have reached out to us and asked questions. On Tuesday the 26th, at 3:00 PM Eastern Time, we will host an open Q&A for anyone that has questions and/or wants to find out more. Use this link to join: https://meet.google.com/gii-aaip-uga.
Open Q&A with the Librar team
Tuesday · 3:00 PM Eastern Time · Google Meet
Thanks,
The Librar team
Questions in the meantime? carl-hugo@librarlabs.com
