6 takeaways from the Charleston Conference — and not just for technical services. This conference, even online only, expanded my awareness on the current depths of some library profession challenges, particularly with AI. (Who can talk about anything without mentioning AI?) I didn’t write up the licensing stuff, but there definitely was some. A lawyer at the University of Michigan has big aspirations, but knows they are practically impossible, so I emailed that maybe he’d want to consider updating NISO’s SERU as a realistic outlet for his good idea: licenses really only need to be 3 pages. (If only they were!)

  • $MONEY$ Please read this online write-up of “Follow the Money to Understand Publishers’ Priorities” in Research Information! I was blown away by Doug Way’s review of vendor’s economics, interests, and areas of profit. I immediately acted on his suggestions to follow him in LinkedIn and sign up for The Brief, the monthly newsletter of Clarke & Esposito (C&E). He is “sour” on OA and Transformative Agreements because of seeing how much money the publishers are making that way. He advocates that librarians “use your voice to speak with vendors and vote with your dollars.” I was stunned to learn that many libraries stayed with Clarivate/ProQuest for e-books, going with the subscription when the vendor dumped the transactional/ ownership (“lumpy” or “spiky”) model and abandoned perpetual access for archival collections. It surprised me to see how much Clarivate’s profits grew from that change in spite of their losing $250k in revenue with the change, cutting costs by $200k (including RIF) as a result. In contrast, Gale reported double digit percentage losses as they continued with the perpetual sale of archival collections — because libraries didn’t do year-end big purchases as much, presumably because dollars were taken up by subscription models. Libraries are accepting these models through inertia according to Doug. I’ve been saying that the consumer is always the loser at the all-you-can-eat buffet ever since Clarivate announced the drop of transactional e-book sales. I was only thinking that you don’t get your money’s worth because most people don’t eat up to the spend level, but I had not thought about the additional impact of other restaurants, with their unique dishes, possibly closing down while everyone goes to the buffet restaurant for the supposedly greater variety. I’m feeling like my analogy is even better than I realized, but I’m saddened by the knowledge.
  • Togetherness vs isolation (“For-Profit v Not-for-Profit Providers of Content and Services: Should it Matter for Librarians?”) The panel agreed up front that it was less about profit vs non-profit than other types of dichotomy that can have impacts. In opposition to my feeling described above about librarians having choices in the marketplace, I’ve always felt that faculty are the main drivers in dictating what journals we must have and thus librarians are not in control of directing the choices there. One librarian on this panel with two publishers from the UK was in favor of librarians taking greater control (ensuring purchasing from smaller publishers, not just the main big commercial ones), but I honestly just can’t see it as a choice. Even if we’re encouraging faculty to take advantage of our transformative agreements to publish their articles in a journal where there’s no perceptible additional cost (and we’ve just obtained a new tool to make that more visible), that’s not the deciding factor (regardless of budget size), is it ? Y’all tell me if you see/hear differently here at WFU, but my impression is that faculty already know what journal(s) they need to try to publish in if they are to really take part in that scholarly conversation for their area of expertise. They have to be reading in a particular journal or two and respond to the colleagues in their discipline in those same channels, right? If you’re not heard, you’re not participating in the conversation, you’re crying (barking?) in the wilderness, right?
  • “Can We Trust ebook Usage Data? An Analysis Across JSTOR, Fulcrum, and Big Ten Open Books” The bottom line is that we can’t really trust our recent usage stats for e-books. I learned from Jeremy Morse that agentic AI and Cloudflare to block bots can inflate/deflate ebook usage stats. He says Liblynx possibly will have a more transparent firewall than Cloudflare (allowing more refined settings). JSTOR: 1) saw ChatGPT traffic go from approximately 5 million to 52 million in 2025; 2) usage reported to libraries is declining overall. COUNTER reports have been focused on human usage, filtering out bots and crawlers. Counter is working on how to include AI-based traffic. Everyone seems to want to figure this out.
  • Librarian-based innovations
    • “Lasso Your ChatBot: Tying Institutional Knowledge to LLMs to Make a Reference Chatbot” (2 librarians from Vanderbilt)
      • Josh: Find the right database for your research — Started with a ChatGPT agent and the Springshare Databases A-Z (LibApps A-Z API), library staff directory page (research and subject roles only); also has an OA journal recommender for authors included in this project. Initially he had had problems with hallucinations (even making up librarians) and it took lots of tweaking, but finally working — see http://tiny.cc/vdb
      • Chuck: wanted to scale down above to business resources — BizTel Assistant (Claude) — “Factiva has a lot of friction in terms of syntax” to search it so wanted BizTel to recommend how to navigate it. He created an LLM Wiki (Wiki Markdown files with a knowledge graph on top of it); chatbot deployed to AWS – Lambda mentioned. It encourages the patron to seek a librarian. He ran an example asking which databases have good consumer info. He’d like to also incorporate Libguide info.
      • It seems to me that what they are doing is taking stuff we already provide through a variety of interfaces and moving it into self-service for patrons via an AI interface.
    • “Getting Repeatable Results from Generative A.I.” Used AI to write first a template, then an agent, to extract key info from long VPATs (and create a score), and this way can add new VPATs on an ongoing basis — no starting over each time, even with a template. The librarian learned that a data dictionary was a key piece in making it all work well. But a key takeaway for me was that the librarian had fully intended to treat Copilot solely as a tool, without humanizing it, and ended up “arguing.” Copilot continued to throw it back at her later with phrases like “so I’m not completely useless to you….”
    • “One AI Workshop Is Not Enough: Building Role-Based AI Literacy Windmills for Library Workers” https://ai-for-academic-libraries.vercel.app/ The following is all lifted from the web site: Learn to use AI critically, responsibly, and practically in academic library work: A free, self-paced curriculum covering AI literacy, research support, reference, metadata, automation, agentic AI, and AI-powered tool building. Start with your role The curriculum covers both paths. Pick the one that matches your work and we’ll highlight your recommended modules across the curriculum – or browse everything.
      • Reference, instruction, research support Practicing Librarian You sit at a reference desk or teach library sessions. You want AI to help you today – drafting emails, building lesson plans, supporting student researchers. Your track: 15 modules — 01–07, 10–16, 18
      • Metadata, cataloging, systems, repositories Digital Librarian You work with systems, metadata, digital collections, and institutional repositories. You want to understand where AI integrates with the tools you manage. Your track: 16 modules — 01–05, 08–18