As the impact of generative artificial intelligence continues to grow, are there ways that libraries can help shape its future?

The growing pace of technological change is difficult to ignore. While it took the telephone 75 years to reach 100 million users, the World Wide Web required only seven to achieve the same milestone, explains the new book Generative AI and Libraries: Claiming Our Place in the Center of a Shared Future, published by ALA Editions and CORE this summer. Coauthors Chris Rosser, first year and transfer experience librarian at Oklahoma State University, and Michael Hanegan, founder of the Center for the Future of Learning and Work and an adjunct professor of artificial intelligence (AI) at Rose State College and the University of Central Oklahoma, note that Open AI’s generative AI tool ChatGPT reached 100 million active users just two months after its public launch.

Aside from AI’s rapid permeation of education and work, it is leading to several surprising advancements. “When ChatGPT launched, I couldn’t have imagined developments like [Google] DeepMind’s AlphaFold, which has accomplished what amounts to a billion years of scientific research in protein modeling,” Rosser tells LJ. “These breakthroughs have shown me that we don’t yet have mental categories for the sheer scope of what’s becoming possible.”

Rosser and Hanegan describe generative AI as an “arrival technology”—like electricity or the internet, for example—that will “fundamentally reshape society regardless of individual choices or adoption.” This certainly echoes the inevitability narrative coming from Silicon Valley, but they make a convincing argument that if libraries want to help shape how people use AI ethically and be part of the broader conversation about AI—even if in a critical manner—then the time to get involved is now. The stakes are simply too high for both the field and the people it serves. “Public libraries are about to be on the front lines of the largest reskilling effort in human history,” Rosser tells LJ. To prepare, he says librarians will need to learn the technology and be able to help teach others; connect with local businesses and workforce development organizations to “identify the competencies our communities most need to develop,” and coordinate to share resources across public, academic, school, and other libraries to amplify the field’s collective impact.

Other librarians share related concerns. “It seems to be more and more likely that we are entering a world—we may already be in a world—where AI is a necessary tool for academic and professional success,” says Dr. Michael Hughes, dean of the Newman Library at Baruch College, City University of New York. “In such a world, the people who really know how to use those tools, and the ones who have access to powerful and current pro-versions of those tools, are going to have a distinct advantage over those who don’t. And that’s going to deepen inequities.” Hughes adds that “librarians, from day one, have been active and critical users of AI. We are really poised as a profession to continue to engage actively with AI and provide timely and useful guidance to our students.”

Even at this relatively early stage, generative AI has been growing and changing so quickly that the idea of libraries asserting any sort of guidance over its direction may seem hopeless. “We are talking about massive global power—some of the biggest corporations in the world, predominately the big five tech companies—that are dominating the AI value chain all the way across the board,” says Alison Macrina, founder and executive director of the Library Freedom Project. “From computer power to datasets to the foundational models, the entire ecosystem is represented by these enormous companies.”

However, Macrina adds that “I think that it is possible for libraries to have an influence on the future of AI, but I think it really depends on our ability and willingness to take a critical posture toward it, and position [libraries] as a place where our communities can get information about how it influences their lives, what its risks and harms are, and ways that we can get organized to reclaim some of our agency and power against the worst parts of it.”

Many libraries already have influence over how AI is taught and adopted in K–12 and academia, Hanegan says, and “here in Oklahoma, we’re having major conversations with our state department of libraries about how we procure for all state agencies. So, there’s huge leverage.” While that leverage is not always direct, “there is a collective power [in libraries] that I think is untouched,” he says. Also, while software engineers are great at building tools, they may have gaps in their skills that libraries could potentially do more to assist AI developers with, such as pedagogy and knowledge management, he says. “Where we have both competencies and scale are some really interesting spaces to push.”

CIVIC DUTY

In addition to the field’s collective power, libraries can have a great deal of influence locally, says R. David Lankes, the Virginia and Charles Bowden Professor of Librarianship at the University of Texas at Austin and cohost of LJ’s Libraries Lead podcast.

“Right now, the place where librarians and libraries could have the most impact isn’t on trying to change OpenAI or Microsoft or Google; it’s really in looking at implementation policy,” Lankes says. For example, “on the public library side, many cities and states are adopting AI policies now, as we speak,” Lankes says. “Where I am in Austin, the city has more or less said, ‘go forth and use AI,’ and that has turned into a mandate for all of the city offices, which in this case includes the Austin Public Library” (APL).

Rather than responding to that mandate by simply deciding how the library would use AI internally, APL created a professional development program to bring its librarians up to speed with the technology so that they can offer other city offices help with ways to use it, and advice on how to use it ethically and appropriately, Lankes explains.

“Cities and counties are wrestling with AI, and this is an absolutely perfect time for libraries to be part of that conversation,” Lankes says. “They need to know what they’re talking about, obviously. There’s a need for basic AI literacy in terms of how it works, but more important is the connection to their communities” and understanding the needs, concerns, and uncertainties about AI in their communities in order to communicate that to other city offices.

And “share, share, share” with other libraries, Lankes advises. “If your city is doing it, some other city has done it.” Check with organizations such as the American Library Association, Urban Libraries Council, or state library associations before attempting to reinvent the wheel. Also, many university libraries have published LibGuides on creating AI policies. Those guides have primarily focused on academic and library system policies, Lankes notes, but libraries could use them, combined with any available examples of civic and municipal policies, to get ideas and possibly craft templates.

Assisting local government with the development of AI policies “is something at the practical level that we can do,” Lankes says. “It’s in our field of finding information, organizing information, and keeping information up to date. And it’s a direct, specific action that we can take that will improve libraries’ position” with the local government and within the local AI conversation.

GATHER, LEARN, CREATE

On a different scale, academic libraries can help influence how their institutions address AI, and can have a significant impact both on academic policies and how students learn to use the technology responsibly.

In August, the University of North Carolina at Chapel Hill Libraries (UNC) launched an AI Studio for students and faculty at its Walter Royal Davis Library. Sharing space with the Library Data Services Desk, the AI Studio will offer collaborative workshops, speaker events, hands-on instruction, and a central location for community dialogue regarding AI. Since most AI solutions are now cloud-based, the new studio didn’t require any specialized computer equipment, notes Tim Shearer, associate university librarian for digital strategies and information technology for UNC, but it will feature personalized support from library staff to help students and faculty use AI tools and build AI solutions.

Michelle Rodell, associate university librarian for health sciences and director of UNC’s Health Sciences Library, says that the studio concept was kickstarted in spring 2024, when UNC’s Vice Provost of Libraries, María Estorino, brought a small group of leaders together, including Rodell and Shearer, to discuss how the libraries would be approaching generative AI in terms of staff and patrons. The group created a roadmap, with the AI Studio proposed as a way to establish a dedicated cross-departmental library space that faculty, researchers, and students could all use.

At a later offsite meeting about generative AI organized by the dean of UNC’s School of Data Science and Society, Stan Ahalt, that brought together all campus leadership currently dealing with AI and the deans of all UNC’s schools, Rodell and Shearer pitched the AI Studio concept, “and it really got a lot of support,” Rodell says. “The main message that we were sending was the democratization of access to tools among our students and faculty.”

Shearer adds that they came to the meeting with “a shovel-ready project…. There were three projects that everyone got behind, and one of them was the AI Studio.” UNC had funded an acceleration program for AI, and as part of that, “the AI Studio has been funded centrally by campus with support from the deans, the provost’s office, and central IT. It has been really nice to have a funded project in the libraries where we’re seen as a leader.”

Gather, learn, and create are the three themes in the new space as it is established. The centralized location is key to the “gather” part of the mission, Rodell says. “We serve everyone…all of campus, so we’re not embedded in a specific school…. One of the things we’re setting out to do is create a space for people to come together to talk about issues related to GenAI, whether that’s ethics and sustainability, or differing opinions on how it should be used in the classroom, or what the student perspective is, or just an opportunity to showcase how our students and researchers are using these tools.” The “learn” theme is captured by the workshops and programs that the library has already been doing, but that will now have their own home on campus. And the “create” theme is a jumping-off point, where the library can help faculty and students bring AI concepts to life.

KEEPING PACE

One of the biggest challenges librarians face with any efforts to influence the future of AI, or even the way their patrons use it, is the speed at which it continues growing and changing.

“We should never think about how ‘the librarian’ is supposed to do this,” Hanegan says. “We should think about how librarianship is supposed to do this. Part of leveraging [the field’s] collective power is that there are people with the ability to stay closer to the frontier in specific spaces. Those people should be understanding, distilling, and disseminating that information in an accessible way for librarians at scale.”

In terms of helping patrons with this rapidly changing technology, Fernando Aragon, learning labs supervisor at Gwinnett County Public Library (GCPL), GA, teaches courses on AI and Game Development at Gwinnett Technical College; Game Art and Programming at Georgia State University; and free coding courses and programs on generative AI at GCPL. For his GCPL program “Generative AI: Separating Fact from Fiction,” he begins with the assumption that attendees could have widely varying levels of knowledge about the technology. “Like preparing for other programs and workshops, having a plan B and C definitely helps for different learning styles,” Aragon says. “Surprisingly, my preparation to discuss AI for the college courses I teach is very similar. AI is constantly evolving, and there may be new AI models or capabilities released semester to semester. So, there is the importance of meeting the learners where they are, and, frankly, learning with them.”

In terms of “learning with them” and keeping up with the rapid developments in AI, Aragon plans to propose monthly discussion groups at GCPL, where patrons can talk about the latest AI news, what they are hearing, and how they are using it. He hopes it will also serve as an opportunity to identify needs in the community that the library can help meet. Aragon is planning to have open discussions once per month in his college courses to help get everyone on the same page with the latest AI developments.

Rosser contends that something comparable is needed for librarians and library staff, noting that libraries need to “create structured opportunities for staff to share discoveries and failures, [and] establish ‘AI literacy’ as an ongoing professional development priority, not a one-time training.”

For research, news, and publications that present a more critical view of AI, Macrina recommends checking out the Distributed AI Research Institute (DAIR) at dair-institute.org and the AI Now Institute at ainowinstitute.org. Macrina also says that independent tech journalism site 404media.co, while not exclusively about AI, “is really staying on top of the AI story, and looking at it from a power analysis [perspective]—who’s behind it? What are their political objectives?” She also suggests the Mystery AI Hype Theater 3000 podcast—back episodes are hosted on DAIR’s site—techwontsave.us (which, like 404media.co, is not exclusively about AI, but covers the topic regularly); tech writer Ed Zitron’s AI news blog wheresyoured.at; tech journalist Brian Merchant’s website and newsletter at bloodinthemachine.com; and mediajustice.org’s series on the “tech broligarchy,” which “really gets into that political and ideological dimension that I think is so important for us to understand, as we understand how AI works.”

In addition, any library workers interested in becoming more active in the library AI space might consider checking out the Library Freedom Project’s AI Working Group, “where we’re building resources and educational materials to help both our patrons and our colleagues understand the AI hype and pushback against some of the uncritical adoption,” Macrina says, adding that interested librarians can get in touch at info@libraryfreedom.org.

Aragon notes that librarians should also be aware of misinformation and misunderstandings about AI. “While it is important to digest scholarly journals and research, also read and watch what is being said from unreliable sources as well,” he says. “Know what your community is hearing about AI regardless of whether it is factual, as the information can be misleading or dangerous, and prepare to counter and educate on that. Also, take the time to use the tools. I like to prompt about things that I know are true, and then see whether or not I get the right answer.”