AI for Librarians — the Training Worth Your CE Budget

Start free, and start this week. RDMLA’s AI for Librarians is a three-hour Canvas course built to the ACRL AI Competencies, taught by working librarians, ending in a certificate of completion. If your library will pay, add Nicole Hennig’s six-week AI Literacy for Library Workers ($99 self-paced on Udemy) or Library Juice Academy’s Practical AI for Librarians ($250, 1.5 CEUs). Don’t pay $129 for a 90-minute webinar.

One test decided the ranking below: does the course train you to judge machine output, or only to produce it?

What AI for librarians actually changes in 2026

The job didn’t get automated. It got inverted.

For twenty years the reference problem was scarcity. Now it’s fabricated abundance: patrons arrive holding citations that read perfectly and don’t exist. 404 Media reported in September 2025 that librarians were fielding requests for hallucinated titles, with volume spiking after an AI-generated summer reading list ran in special editions of the Chicago Sun-Times and The Philadelphia Inquirer. Coverage of that reporting puts the Library of Virginia’s estimate at roughly 15% of emailed inquiries now originating in AI misinformation (verify: secondary relays, not the paywalled original).

The newly valuable competency isn’t writing a clever prompt. It’s running a fast existence check on a plausible citation, then explaining the result to someone convinced you’re hiding the book. Some take it badly, and almost no general AI course prepares you for that conversation.

The profession has now written this down. ACRL’s Board approved AI Competencies for Academic Library Workers on 3 October 2025, organized around ethical considerations, knowledge and understanding, analysis and evaluation, and use and application. It’s the document a director points at when they ask what training you need. Courses that map to it clear a budget line. Courses that don’t have to argue for themselves first.

Where AI in libraries has already landed

Discovery grew a summarizing layer you didn’t choose. Ex Libris documents the Primo Research Assistant pipeline plainly: up to 30 retrieved results get re-ranked down to five, and the overview a student reads is written by GPT-4.1 Mini from those five abstracts and nothing else. Ex Libris even concedes in its own documentation that the overview “may not utilize all top five sources.” EBSCO’s AI Insights compresses a full-text article into two to five key points inside EBSCOhost and Discovery Service. The one-shot that used to teach Boolean logic now has to teach why the assistant kept those five sources and what it quietly dropped.

Cataloging went next. AI-assisted MARC and Dublin Core generation, authority-file remediation, backlog cleanup: all useful, all wholly dependent on someone who can spot a confidently wrong subject heading. It’s the most employable AI skill in technical services.

Then there’s internal paperwork, where most librarians actually recover hours. Grant narratives, board reports, program blurbs, policy drafts, ILS exports. Unglamorous, immediate, and nobody builds a webinar around it.

Selection and policy come last and matter most. Vendor platforms ship AI features on by default, raising patron-privacy questions your license may not answer, and machine-generated books now arrive in vendor catalogs with clean metadata and no real author. Generative AI in libraries is a policy problem before it’s a tools problem, which is why the good courses make you produce a document instead of collecting a badge.

The AI courses for librarians worth taking

Prices below come from the provider’s own page, checked 5–6 August 2026. Where a page wouldn’t render us a price, we say so rather than guess.

1. RDMLA’s AI for Librarians: free, and take it first regardless

Free · ~3 hours · Canvas · certificate of completion

The Research Data Management Librarian Academy built it, co-led by Elaine Martin of Harvard Medical School’s Countway Library and Rong Tang of Simmons University. It maps to the ACRL AI Competencies, and it’s taught in 10–18 minute segments by working librarians the team recruited through an open call for proposals. Three units: AI tools for library research (chatbots, Perplexity, Elicit, Claude, NotebookLM), AI ethics, and use cases across administration, reference, research data management, and instruction. Nicole Hennig leads Unit 1. Nearly a thousand people took it in the first six weeks after its late-2025 launch, and a fourth unit on metadata and cataloging is planned.

Where it falls short: it’s academic-library-shaped. Public and school library staff will find the research-data framing off-center, and three hours of video makes you literate, not fluent.

2. AI Literacy for Library Workers (Nicole Hennig), the hands-on one

$99 self-paced on Udemy · $200 via an Infopeople cohort · 6 weeks

Six weeks that make you build things rather than watch things: a specialized chatbot, alt text via computer vision, a generated podcast from your own documents, bias tests on image generators. The weeks run technology, prompting, ethics (copyright, labor, climate), multimodal features, multimedia creation, and the sleeper of the set: how to stay current without swallowing the hype.

The catch: the Infopeople cohort page still advertises an October 2025 start, so the live version runs irregularly. Take the Udemy route unless a cohort is confirmed open, and budget six real weeks; past students call it time-consuming.

3. Practical AI for Librarians (Library Juice Academy) files the CEUs

$250 · 4 weeks · ~15 hours · 1.5 CEUs or 15 PDHs

Taught by Denielle Wells, a librarian with 17 years in the field; next session 5 October to 1 November 2026. ChatGPT, Perplexity, and Canva Magic Write run against metadata creation, reference support, outreach, and programming, with privacy, transparency, and bias threaded through. You finish holding a deliverable: a draft policy, workflow, or literacy program for your own library. Most general AI courses give you nothing to file with your board.

It sits inside LJA’s Certificate in Artificial Intelligence in Libraries: four required courses (this one, AI and Libraries, for Skeptics, the Deep Dive sequel, and Algorithms Can’t Shelve Empathy) plus two of four electives, including Introduction to AI and Metadata for technical services and AI for Library Instruction for academic and school staff. All eight are $250.

What you’re really paying: $250 for fifteen hours is the steepest hourly rate on this page, and the full certificate means six courses at $250 each. Call it $1,500. That’s our arithmetic, not a published figure: Library Juice publishes no program price at all. Don’t count on shaving it, either. The advertised 10% is for registering for every course in some of its certificate programs, and this one asks you to pick six of eight.

And the credential carries no accreditation. Library Juice discloses that itself on its certificates page, calling the programs “not accredited by an outside institution” before adding that they are “widely recognized as a marker of significant knowledge and advanced skills.” Inside libraries, that recognition is real. Outside them, nobody will know what it is.

Alternative: Library Journal’s live course, for training a whole team

$256 early bird / $303 standard · half-day live + 3 self-paced modules · 12 PD credit hours · certificate

Crystal Trice teaches “AI for Librarians Who Don’t Have Time for This” on 10 September 2026 (noon–4pm ET), repeating 4 March 2027 at $249/$299. The $256 early-bird rate closes 10 August 2026; after that it’s $303. Best for the reluctant: evaluation for accuracy, bias, copyright and environmental cost, plus permission to decide what’s “worth saying no to,” in LJ’s own phrasing. Recorded, six months of access, and multi-seat discounts calculate in the cart, which is its real use case.

The trade: more money than Library Juice for fewer filed hours. You’re paying for four live hours with a consultant, not a bigger number on your CE form.

General base: Google AI Essentials

Five short courses, under 10 hours, free to enroll, and about 1.9 million people have. Coursera showed 24,664 reviews across the program on 6 August 2026. Not library-specific, which is the point. The library courses all assume you can already talk to a model competently. Our full write-up: Google AI Essentials review.

Courses compared

Figures read on the providers’ own pages on 5–6 August 2026, with two exceptions: Elements of AI’s terms carry from our 30 July 2026 pass, and Hennig’s $99 comes from nicolehennig.com because Udemy’s page renders its price client-side and gave us nothing.

Don’t bother with the $129 webinar certificate

Here’s the clearest waste of money in this category.

Library 2.0 sells “Customizing AI,” a 90-minute workshop with Reed Hepler, at $129 per person, including the recording, slides, and a participation certificate. Its ChatGPT & AI Bootcamp, three one-hour sessions, is $149. On the same site, “Libraries and AI” is also 90 minutes, also carries a participation certificate, also includes the permanent recording and slides, and is free. (All three prices read on library20.com on 5 August 2026; the site was down for maintenance when we rechecked on 6 August.)

Check the dates before you pay. That $129 session went out live on 23 September 2025, so what’s on sale now is a recording priced like an event. The $149 bootcamp is worse: its three sessions ran in May and June 2024, and the page openly says registration now buys immediate access to the recordings. Group rates soften the $129 ($99 each at 3+, $75 at 5+), which mostly proves the list price is soft.

For the same $129 you get two of the four weeks of a Library Juice course built on fifteen hours of graded work — or, free, the whole RDMLA course with a certificate. Buy the one-off webinar only when the topic exists nowhere else and your employer pays. The same caution applies to the hour-long AI recordings in ALA’s eLearning catalog.

The genuinely free path

Do RDMLA’s three hours first; it’s the only free course here built to the profession’s own competency document. Then make an account at learn.webjunction.org, a program of OCLC that is free to every library “regardless of size, type, or location.” Finish anything in its catalog, self-paced course or webinar recording, and you can print a certificate of completion from the course page.

Check vintage as you go. The much-cited “AI and Libraries: Enhancing Services and Engaging Communities” is a webinar recording from 9 August 2023 — the framing holds, the tool list doesn’t. Watch “Where is AI in 2026, and where is it going?” instead, recorded 21 May 2026 with Andres Ramirez of the AI Safety Awareness Project, which spends real time on what these systems are bad at.

One caveat WebJunction states itself: whether any of it counts for CE or LEU credit is decided by your local agency, not by OCLC. Confirm before you build a CE plan on it.

Add Library 2.0’s free “Libraries and AI” webinar for sector framing, and Elements of AI if you want conceptual grounding with a certificate that costs nothing. Then check your state library’s continuing-education catalog. That’s often where the CE credit your board wants actually lives, free to in-state staff.

Say you become the AI person in your system: writing the policy, training the staff, answering the board. The sector courses run out of depth fast at that point. Our free AI automation course builds the wiring the library courses describe but never construct, and Towards AI runs hands-on training with a reader discount. One honest line: exhaust the free path and the $250 courses first, then upgrade at their ceiling.

The full no-cost route is mapped on our free AI courses hub, and the AI certifications guide ranks credentials by what employers recognize. Hiring for an emerging-technologies or digital-scholarship role, or looking for one? See the talent board. This page sits under AI courses by profession.

FAQ

What is the best AI course for librarians in 2026?

Start with RDMLA’s AI for Librarians: free, about three hours, built to the ACRL AI Competencies, with a certificate. For hands-on depth, Nicole Hennig’s six-week course is $99 self-paced on Udemy. Pay Library Juice Academy’s $250 only when you need documented CEUs or a policy deliverable.

Is there free AI training for librarians?

Yes, and it’s good. RDMLA’s AI for Librarians is free on Canvas with a certificate. WebJunction, a program of OCLC free to every library, offers library-specific courses and archived webinars that each earn a certificate of completion. Library 2.0’s “Libraries and AI” webinar is free, and Elements of AI’s introductory certificate costs nothing.

Should I get the full Certificate in Artificial Intelligence in Libraries?

Only if someone else is paying. It requires four courses plus two of four electives at $250 each, about $1,500 by our arithmetic, since Library Juice publishes no program price. Two well-chosen courses deliver most of the value. Pursue the full certificate when building a formal technical-services or emerging-technologies specialty.

How are librarians using ChatGPT day to day?

The highest-value uses of ChatGPT for librarians are internal: grant narratives, board reports, program descriptions, policy first drafts, and cleaning up ILS data exports. Patron-facing use needs far more caution, since fabricated citations and invented titles are now a routine reference problem.

Will AI replace librarians?

The opposite pressure is showing up. Machine-generated citations, fake book requests, and AI-written titles entering vendor catalogs all raise demand for people who verify claims and teach others to do it. Routine lookup work shrinks. Authentication, instruction, and policy work grows.