Take a Look at Choice’s Webinar Polling on AI in Academic Libraries

Our data on AI use, policy, and confidence

Posted on March 25, 2026 in Blog Posts

Choice runs numerous webinars throughout the year, covering a diverse set of topics relevant to academic librarians. As a part of these webinars, presenters sometimes run quick polls of the audience members to gauge their feelings and experience levels. Outside of the webinars, we have never released these data, but we felt that they may be of interest to the broader academic library community. Below, we walk through what several of our polls show about AI in academic libraries.

Of course, a few points of caution should be kept in mind:

• Our audience is, naturally, self-selecting to a degree. Not every person who attends our webinars or votes in the polls is an academic librarian, though most are.
• Similarly, most who attend our webinars have a stake in the topic. Experts are less likely to attend an informative webinar on a topic they’re well-experienced in. People with a fundamental opposition to AI aren’t likely to attend, either.
• The raw data have been processed into readable graphs, but beyond that, they haven’t been refined to reflect institutional types, professional positions, and so on.

In other words, these polls shouldn’t be taken as rigorous studies of academic libraries, and I have inserted caveats where I think appropriate. What they do provide, however, is a snapshot worth mulling over a cup of coffee.

🧑‍🏫 Check out Choice’s upcoming webinars

---

We wanted to begin by looking at our earliest poll data on AI. The above findings were taken from Leo Lo’s LibTech Insights webinar in September 2023, “CLEARer Dialogues with AI,” on the topic of prompt engineering. Though the webinar itself remains evergreen, the poll data provide a snapshot of what the AI landscape looked like in late 2023, about a year after ChatGPT’s public debut: 79 percent of the audience had used ChatGPT before, and 21 percent had not. These data aren’t altogether surprising, given the topic, but as we can see, a sizeable minority hadn’t used ChatGPT before.

It’s interesting to compare these data to the latest Pew Research findings on AI use among adults. In late 2025, Pew found that 62 percent of US adults interact with AI at least several times a week. The question of whether or not one’s tried an AI tool is outdated (as is the exclusive focus on ChatGPT), but this poll is a good artifact of where the profession was in 2023.

---

🔥 Stay up-to-date with LibTech Insights by signing up for our free newsletter. Just one weekly email with our new blog posts, top tech news stories, and other bonus content. Check out some posts from our archive:

• Why Your Library Needs a Style Guide
• Benchmarking AI Models in the Library: Critical Questions for Library Professionals and Users
• Maximizing Impact: Using Excel Solver for Optimal Library Budgeting

💜 ALA is having its 150th anniversary. Come celebrate with us!

---

Compare the above data with more recent data taken from our November 2025 webinar, “AI Meets Evidence: Practical Applications for Librarians in the Review Workflow.” We see a normal distribution of responses, with most falling somewhere in the middle in terms of familiarity with AI and LLMs.

These data provide an important rebuttal to the binary that we often see in discussions on AI use. Despite the all-or-nothing framing that one must either want to accelerate to the AI-governed future or refuse AI altogether, more people likely fall in between these points than at either end. Given the topic of the webinar, it perhaps isn’t surprising that so many respondents fell into the “occasional use” column: many people may be uncomfortable using AI for everything, but they are receptive to using it occasionally for practical applications, hence their decision to sign up for the webinar in the first place.

The caveat I would issue about these data is that the answers switch from “understanding” to “use,” which aren’t necessarily the same. That is, one can have more than a basic understanding of AI and still choose not to use it, or one can use it all the time without understanding it. As I said at the outset, these polls shouldn’t be taken as more than what they are—mere snapshots—and this poll gives us a more nuanced and up-to-date picture of what AI knowledge/use looks like three years on.

In February of this year, we ran a webinar on “Harnessing AI as a Collaborative Partner for Ethical Research and Writing.” The presenters used polls at the start and end of their presentation to gauge the audience’s confidence in helping others use AI tools ethically in academic writing. The graph above shows the audience’s responses at the start of the presentation, and, naturally, most who attended the webinar to learn about the topic were not “very confident” going in.

So, again, I need to issue that caveat about self-selection, but what I find productive about this chart, especially in light of the previous one, is that it adds another layer of complexity. As we think about AI in the profession, we need to consider and measure not only understanding and use but also confidence in leading others through this technological and ethical morass. In our discussions of AI use and implementation, we need to keep all three prongs in mind, because it isn’t sufficient to know and use AI if we can’t speak to others about it confidently.

Our most recent webinar, “AI, Algorithms, and Librarians: The Evolution of the Librarian in the GenAI Era,” offered an important snapshot of the institutional landscape. A limiting factor in the above question about librarian confidence is that many librarians work in institutions without clear AI policies to guide ethical judgment. While university administrations are eager to advertise AI adoption, institutional policy is lagging. Tellingly, only 27 percent of respondents reported that their institutions have an approved policy in place. The caveat is that these data may represent multiple librarians from the same institution, but I think the overall picture likely holds: most institutions are in the developmental stages of AI policy.

This reinforces LibTech Insights‘s position since its inception: as information professionals, librarians have an important role to play in institutional policy development and should be prepared to enter into these conversations. This crucial window is still open, and if possible, librarians should take advantage of this opportunity.

With the same caveat issued above about redundant institutional responses, this graph offers a good place to end this tour. Less than a quarter of libraries currently subscribe to any AI tools. Most have investigated but haven’t made any subscription decisions. These data suggest that libraries potentially have a lot of leverage with vendors right now because most aren’t locked into products or contracts. (Check out Elizabeth Szkirpan’s guide to talking to vendors about AI tools.)

This chart, like the others in this post, shows uneven terrain in the library landscape. Understandably, many people are fatigued or pessimistic about AI, but we are at a moment when important decisions with long-term consequences are being made. Resignation under these circumstances is dangerous when we need confident leaders with both knowledge and values to make important choices about policy and education. Hopefully, these charts will inspire you by showing the work that still needs to be done.

Related Posts