UVA’s AI Literacy Lab Puts Librarians In The Coach’s Seat

The University of Virginia opened a new front in the campus AI fight on April 17. It put a librarian in charge.

A joint venture between UVA Library and the College and Graduate School of Arts & Sciences, the AI Literacy and Action Lab launched with four faculty pilots, a five-competency curriculum, and a model that pairs every professor with a trained librarian coach for an entire semester. Leo S. Lo, UVA’s University Librarian and Dean of Libraries, runs it with Christa Acampora, dean of the College. Classes are already in session this spring across economics and English classrooms in Charlottesville.

At most universities, the AI strategy lives inside a computer science institute. UVA’s lives inside the library.

Why A Library Is Running UVA’s AI Lab

Lo dates the lab’s origin to December 2022. Students at his then-institution had started asking librarians for books and articles that did not exist. The titles sounded plausible. The authors looked real. None of them were.

“Those students needed a librarian before they needed a better AI tool,” Lo said in a UVA Today Q&A on what a library does in the AI age.

He calls the new dynamic an “AI answer economy.” For most of research’s history, finding information meant chasing a citation, comparing two accounts, deciding which author to trust. AI collapses all of that into a single synthesized answer, and the lab is designed to teach students to run the work in reverse.

Acampora said anchoring the work at the library was deliberate, since every school and discipline already touches it. Lo went further. The lab’s premise is that AI literacy is a liberal arts problem as much as a technical one, and a university that treats it only as technical will get the answer wrong.

Inside Lo’s Five-Competency Framework

The lab’s curriculum runs on a framework Lo published as a 2025 preprint titled “AI Literacy for All: A Universal Framework,” now adopted by peer institutions across the country. It pairs with the October 2025 ACRL AI Competencies for Academic Library Workers, the field’s first formal standard for the profession. The framework slots AI capability into five linked competencies, each scored across four progression levels from basic awareness to strategic insight:

• Technical Knowledge. How models are built, trained, and constrained.
• Ethical Awareness. Bias, consent, intellectual property, attribution.
• Critical Thinking. Spotting fabricated outputs, comparing sources, validating claims.
• Practical Use. Workflow design, prompting, tool selection.
• Societal Impact. Labor, inequality, democracy, environmental cost.

Lo argues no single department on a research campus owns all five layers at once. Computer science teaches the technical. Philosophy carries the ethics. Sociology covers the impact. The library, in his framing, is one of the few places “built to convene across disciplines.”

Four Pilots, From Code Sprints To High School Lesson Plans

The lab opened with four faculty pilots that look almost nothing alike, by design.

1. Economics and the Future of Work. Anton Korinek, the UVA economist named to the 2025 TIME100 AI list for his work on AGI and economic transformation, co-teaches with three librarians: Erich Purpur, Meridith Wolnick, and Rebecca Coleman. Students code with Claude Code, Google Antigravity, and OpenAI Codex, then turn to the framework’s ethics and critical-thinking dimensions for group projects on AI’s impact on jobs and inequality.
2. High School Lesson Plans. Piers Gelly’s first-year writing students are designing AI-informed lesson plans for Tim Klobuchar’s AP English and English 11 classes at Charlottesville’s Monticello High School. Facilitators Meridith Wolnick and Sherri Brown anchor a session on critical thinking and ethics.
3. Philosophy and AI. Fall 2026, led by David Danks, the Polk JSF Distinguished University Professor of Philosophy, Artificial Intelligence, and Data Science. Students will use AI to build projects that apply philosophical arguments to real AI deployments.
4. AI-Scaffolded Biochemistry. Andreas Gahlmann, associate professor of Chemistry and Molecular Physiology and Biological Physics, pairs with the lab in fall 2026 in CHEM 4410 to train teaching assistants alongside librarian partners, giving the lab its first STEM template.

A fifth engagement runs this summer through UVA’s AI4VA program, building asynchronous learning modules and weekly office hours for students outside the four pilots.

Every pilot produces a published case study. Faculty get assessment data on how student thinking evolved across the five competencies. Students leave with portfolio artifacts (lesson plans, working code, original analysis) they can show employers.

Librarians As AI Coaches, Not Tour Guides

The model rejects the workshop format that has dominated campus AI training since 2023.

“We deliberately chose the word ‘action’ because people develop AI literacy not by attending workshops about AI, but by using the technology to advance work they care about and reflecting on what happened,” Lo said in the lab’s April 17 launch announcement.

Every pilot has a librarian trained across all five competencies who works alongside the faculty member for an entire semester, not just a one-time guest lecture. Meridith Wolnick, Director of Teaching and Learning at the Library, runs both the economics and English pilots, which means the same person carries the framework across courses in two different schools.

What The Job Market Says About The Bet

The lab arrives at a moment when employer demand for AI capability is shifting fast and student behavior is shifting faster.

Handshake’s 2026 workforce outlook reports that more than 80 percent of rising seniors have used generative AI tools, with about three-quarters describing themselves as “reasonably skilled” or “very skilled.” The share of internship listings mentioning AI skills has grown more than 4x since 2023. Full-time job descriptions referencing AI have grown nearly 5x in the same window.

Anxiety is the flip side. AI’s compression of early-career engineering work has already reshaped how investors evaluate startups, and students feel the same compression. Of the 61 percent of Class of 2026 students who report career pessimism, almost half cite AI as a contributing factor. A separate EAB report found 42 percent of college-bound students say AI will influence their career choice, and 10 percent have already changed their planned major because of it.

A National Network Already Forming Around The Framework

Lo’s role extends past Charlottesville. In the UVA Today Q&A, he disclosed a “statement of shared practice” he has initiated for UVA Library and partner institutions, including Duke University, Oklahoma State University, and Northwestern University, addressing how research libraries respond to AI companies seeking access to archival material for model training.

This matters because the open web is tightening. Publishers are suing or licensing. AI labs that can no longer scrape unrestricted are turning to library archives, and Lo’s collaborative response gives those libraries a shared answer rather than four different ones.

Programming at the new lab runs parallel to that work. Every pilot case study feeds a national evidence base other libraries and universities can build on, countering what Lo calls the “one-off workshop” tendency at most campus AI centers.

Lauren Dodd, director of collection management at Bryn Mawr College, has framed the work similarly, telling Inside Higher Ed that “AI literacy is just a new dimension of information literacy.” Bryn Mawr has positioned its libraries as AI sandboxes for experimentation and consultation. UVA’s version adds the formal structure and published case-study output that lighter sandbox models often skip.

The Honor Code Question And The License Test Lab

Beyond teaching, the lab takes on two functions that often go under-discussed. It will advise UVA on which AI licenses to procure by testing software at small scale before campus-wide commitments. Lo said the lab can pilot tools at faculty or student request, then feed back what worked into procurement. That’s quiet but meaningful plumbing: universities spend millions on enterprise licenses, and most decisions get made without classroom-level evidence.

Coordination with the UVA Honor Committee on academic integrity is the other piece. Faculty are now writing AI clauses into syllabi without a shared institutional framework for what counts as acceptable use. In the UVA Today Q&A, Lo gave one of the more unexpected answers from a dean of libraries: he said banning AI in a particular course can be defensible, then asked those faculty to “stay open” and revisit the question as the lab’s case studies come in.

The lab itself was born of a UVA working group. Interim Provost Brie Gertler convened the body, drawing in the Office of the Executive Vice President and Provost and the Center for Teaching Excellence to advise on AI in teaching, learning, and research. Lo wanted the lab to function as the committee’s “idea and data-generating hub.” Adjacent education-AI moves are accelerating: the National Catholic Educational Association launched free Google AI training across U.S. Catholic schools in May 2026, with corporate-funded initiatives reaching K-12 faster than universities. UVA’s combination of framework, librarian coaches, pilot pipeline, and licensing oversight gives its model an institutional spine that survives leadership cycles.

Frequently Asked Questions

How Can UVA Faculty Apply To Join An AI Literacy And Action Lab Pilot?

Faculty submit a course, assignment, or research project for the lab to map against Lo’s five-competency framework and match with a trained librarian facilitator. Submissions are reviewed on a rolling basis, with pilots running a full semester and pre- and post-assessment built in. Contact Leo Lo at leolo@virginia.edu for the lab’s intake form, or reach Bill Ashby (wla4f@virginia.edu), Senior Associate Dean for Arts & Sciences, for College-side pilots.

When Will The Lab Open To Schools Beyond The College?

Lo has said the College partnership is the launch beachhead, not the ceiling. The library serves every UVA school, so faculty and students from Engineering, Darden, Nursing, Education, Law, and Public Policy can already request pilot facilitation. The fall 2026 series of three one-credit courses spans Humanities, Astronomy, and Music. Expansion beyond UVA is on the roadmap once published case studies validate the model.

Where Can I Read The Full AI Literacy Framework?

The framework lives in Lo’s 2025 preprint, “AI Literacy for All: A Universal Framework,” hosted open access on the University of New Mexico Digital Repository. Its companion is the ACRL “AI Competencies for Academic Library Workers,” released October 2025 by the Association of College and Research Libraries. Both are freely downloadable and useful for faculty or library workers planning similar programs at their own institutions.

What AI Tools Are UVA Students Actually Using In The Pilots?

Pilot one uses Claude Code (Anthropic’s coding agent), Google Antigravity, and OpenAI Codex for hands-on coding work in Anton Korinek’s economics course. Pilot two focuses less on tool training and more on critical and ethical evaluation of outputs across general-purpose chatbots in Piers Gelly’s writing seminar. The lab is also evaluating higher-ed-oriented platforms like BoodleBox used at peer institutions including Bryn Mawr College. Tool selection changes as the lab tests each option.

The bigger story under the launch is who gets to define AI literacy on American campuses. UVA is wagering that the answer lives in a research library running a five-competency curriculum, rather than a computer science institute running a workshop series. The first data on whether that bet pays off arrives when the spring pilots present in April and the fall pilots open in September.