New study: Libraries cut manual workflow time by 30 to 60% with Academic AI
There is no shortage of conversation about the potential of AI. What has been harder to find is evidence of its real-world impact in libraries. The Academic AI Impact Study, conducted by Emerging Strategy on behalf of Clarivate, aims to address that gap and encourage conversation about what effective adoption of artificial intelligence (AI) in library workflows can look like.
The report draws on in-depth interviews with 11 library professionals across eight institutions in North America, Latin America, and the Middle East, examining the impact of AI on two core workflows: course reading list preparation using Leganto Syllabus Assistant, and metadata creation and cataloguing using Alma AI Metadata Assistant. The findings are based on self-reported time comparisons and workflow analysis.
Reported gains
Academic libraries are under sustained pressure to deliver timely, high-quality services even as staffing levels and institutional resources remain constrained. AI is increasingly being introduced in this context, embedded in daily operations, while keeping professional judgment in the hands of librarians.
Across the institutions interviewed, the use of AI was associated with measurable gains in four areas:
Time and efficiency
Time spent on manual, repetitive tasks fell by 30 to 60%. Course list creation, which previously took 15–45 minutes per list, now takes just 2–5 minutes. At the University of Windsor, for example, a 20-item list that previously took 20 minutes to create now takes only 3 minutes.
Throughput and capacity
Libraries reported a 2 to 4 times increase in feasible workload without additional staff. Metadata teams running large enrichment projects processed hundreds of records per day. Universidad Tecnológica de Bolívar (UTB) was able to start processing a cataloguing backlog that had previously been considered unmanageable.
For Dora Lilia Sepúlveda Sepúlveda, Library and Archive Director, Universidad Tecnológica de Bolívar:
“AI is the engine of a historic transformation. By eliminating the bottleneck of manual transcription, we have recovered up to 80% of operational time. This enables us to rescue thousands of long-forgotten records, ensuring that knowledge is visible, accessible, and ready for future standards.”
Access and availability
50 to 60% of reading lists became immediately available to students after AI processing, because library-held materials could be surfaced and linked automatically.
Quality with oversight
70 to 90% of AI-generated metadata was accepted with minor edits. Librarians reviewed every record; the AI produced a first draft, and librarians made the final call.
Amit Niv, Head of Metadata and Process Management at the University of Haifa, described the impact on cataloguing:
“When it comes to cataloguing, the main benefit of AI is in automating repetitive and time-consuming work. It is also a real helper when cataloguing books in languages the cataloger is not proficient in. When using AI to assign subject headings, while you can’t forgo reviewing and sometimes changing them, AI gives you a good starting point and reduces the mental burden of subject analysis.”
What AI changes — and what it does not
Interviewees were consistent on this point. AI changes the operational constraints: it eliminates manual entry as the bottleneck, makes previously deferred work feasible and improves consistency. What it does not change: professional judgment and accountability, baseline staffing expectations, and the need for review, correction and governance. Staff effort shifts from execution to planning, validating and communicating — allowing libraries to allocate more time toward high-value work and engagement with students, faculty and researchers.
Implications for library leaders
AI is producing tangible operational benefits in the libraries that have adopted it. These benefits are not automatic, and they do not look the same everywhere. For library leaders, the implications are about how operational pressure, professional judgment and strategic priorities intersect.
The operational case for adoption
- Libraries in this study adopted AI because existing workflows could no longer absorb current demand. The primary goal was to recover capacity lost to manual, time-intensive processes, not to reduce staff.
- The risk of inaction is reflected in metadata workflows: large volumes of legacy records and uneven descriptive quality limit discovery and readiness for emerging standards and collaborative initiatives.
- For library leaders, the central question is whether existing workflows can continue to absorb growing demands without tools that reduce friction and expand capacity.
Making adoption work
- Outcomes depend on how AI is introduced and governed. Libraries that realized value fastest focused on high-friction workflows where repetitive entry or transcription consumed disproportionate time, rather than attempting broad deployment.
- Treating AI outputs as drafts, with librarians owning standards and the final record, preserves quality and professional trust.
- Adoption was most effective when introduced incrementally, allowing staff to build confidence and integrate AI into existing workflows without disrupting core services.
Read the full study
The full report includes four institutional case studies, each showing a different aspect of how AI played out in practice — from recovering course readiness during a delayed launch, to making large-scale retrospective cataloguing feasible for the first time.
Related insights
The latest news, technologies, and resources from our team.
