Abstract
In 2024, Northern Sydney Local Health District (NSLHD) Libraries implemented a librarian led program to introduce artificial intelligence (AI) tools for evidence retrieval and research workflows to clinicians and hospital staff. The initiative focussed on learning, testing, and teaching safe and appropriate AI use within a health service environment. In 2025, the program expanded to include advanced training and clinical collaborations. This article describes the program’s origins, key design decisions, and developed resources. It also examines early impacts and lessons learned relevant to health library practice.
Why we started
As generative AI tools rapidly entered clinical and research settings, NSLHD Libraries identified the need to build internal capability before supporting organisation-wide use. Librarians required time and space to explore AI technologies, assess their viability, and critically evaluate emerging tools within a health service context.
An AI Community of Practice (CoP) for librarians was established in early 2024. Meeting fortnightly, the group tested tools, shared insights, and discussed implications for professional roles and service delivery. Core aims included identifying potentially useful tools, articulating benefits and limitations, and developing practical guidance grounded in librarianship expertise.
Alongside this work, the library started on an AI education working plan, with defined objectives and timelines developed. From the outset, the AI Essentials program aligned with NSW Health governance and Local Health District Digital Health Service (DHS) messaging, including recommendations to use Microsoft Copilot under enterprise protections.
Designing the program
The AI Essentials program was launched as a series of short, stand‑alone lunchtime sessions offered to the full health district and recorded for later viewing. Initial topics included ‘Chatbots in Action’, ‘Prompting Techniques for Clinicians’, and ‘Find Literature Fast: AI Tools for Quick Results’. In 2025, additional sessions addressed ‘Systematic Reviews in the Age of AI’, ‘Visualising the Literature with AI Tools’, ‘Writing Titles and Abstracts with AI’, and ‘Evidence to Action: OpenEvidence AI, UpToDate, and BMJ Best Practice’.
Sessions focussed on both foundational and applied topics, including ethics and bias, effective prompting, hallucinations and verification, attribution and copyright, and the role of libraries in supporting evidence‑based workflows. Emphasis was placed on responsible use and professional judgement rather than promotion of specific tools.
An AI Essentials intranet page supported the program by providing access to session recordings, AI subject guides, and a clear contact point for questions and requests.
Launch and delivery
The initial sessions were delivered between October and November 2024. Live demonstrations explained differences between large language models and grounded tools, and illustrated verification workflows such as tracing citations to confirm accuracy.
Across all sessions, consistent messages were reinforced: open AI tools may not retrieve paywalled literature; AI assisted searches are not reproducible in the same way as database strategies; and AI generated outputs must be verified and appropriately acknowledged. Privacy, transparency, and policy alignment were embedded throughout delivery.
Living subject guides
Two complementary subject guides, AI in Healthcare and AI in Research, were developed as living resources. Available via the AI Essentials intranet page and the library website, the guides support 24/7 self-directed learning.
The AI in Healthcare guide focuses on safe and responsible AI adoption in clinical and education contexts, including governance and privacy considerations. The AI in Research guide supports research workflows such as prompting, discovery, verification, citation, and attribution. Both guides are reviewed quarterly and updated in response to tool developments, policy changes, and staff feedback. Recent updates have included material addressing environmental concerns related to AI use, ambient scribes, and the use of AI in the mental health arena.
Tailoring for clinical audiences
In 2025, growing demand led to the extension of the program into clinical in‑service contexts. Tailored sessions were delivered to mental health nursing teams, the Aboriginal and Torres Strait Islander Health Service, allied health specialities, and primary and community health departments.
These sessions were adapted to professional contexts and incorporated published literature demonstrating current AI use within relevant specialties. Feedback reflected interest alongside caution, with clinicians appreciating balanced discussion and specialty‑specific examples. Short, practical sessions with live demonstrations lowered barriers to participation. Librarian‑led framing positioned the library as a trusted guide, while recordings, subject guides, and a central hub supported consistent messaging and ongoing access.
Reach, Feedback, and Impact
The AI Essentials program demonstrated strong early engagement, with over 100 participants attending live sessions in 2024 and sustained interest through recorded content, which was viewed more than 400 times that year. The associated intranet page similarly recorded over 400 views, highlighting ongoing staff demand for practical AI guidance. Complementing this, the Library’s AI in Healthcare and AI in Research guides are among the most highly used across the library’s online guides, indicating broader and sustained engagement with AI‑related resources.
Post‑session feedback suggested the sessions met participant expectations and were relevant to daily practice. Participants particularly valued practical demonstrations and clear discussion of both opportunities and risks associated with AI use. Participants in AI library education reported increased confidence in prompting, evaluating AI tools, and checking citations.
The Chief Executive of the NSLHD personally invited a librarian to be among three speakers on a Town Hall style panel on Artificial Intelligence in Health. Attended by 129 people live, and 533 views of the recording within the following fortnight, this reinforces that librarians are now recognised authorities on AI within our Health District and provided an opportunity for greater reach of our AI education.
One of the most rewarding aspects of this work has been seeing clinicians and staff grow in confidence as they engage with AI tools. Librarian‑led education has played a key role in demystifying AI, supporting informed decision‑making, and building trust during a period of rapid technological change. These experiences have reinforced the evolving role of hospital librarians as educators and advocates for responsible AI use in healthcare.
Future directions
Ongoing AI education at NSLHD will continue to evolve, with plans including additional sessions in 2026 on practical Copilot uses, critical appraisal, finding and interpreting evidence.
Strong alignment with governance and safety‑by‑design principles will remain central, alongside closer collaboration with Digital Health Services, Communities of Practice, and clinician partners. Continued engagement with the HoLP NSW Hospital Librarian network will support shared learning and sector‑wide capability building.
Conclusion
The rapid emergence of generative AI has created both opportunity and uncertainty for healthcare organisations. At NSLHD Libraries, a librarian‑led approach to AI education enabled staff to engage with new technologies in a way that was practical, policy‑aligned, and grounded in professional values. By prioritising internal capability building, governance awareness, and verification‑first workflows, the AI Essentials program positioned the library as a trusted partner in supporting AI use across clinical research contexts.
Key elements of success included short, focused training sessions, reusable resources such as living subject guides, and a deliberate emphasis on ethical and transparent application rather than tool promotion. While increasing demand and rapid technological change presented challenges, these constraints highlighted the importance of continuous review and cautious guidance.
This experience demonstrates how hospital librarians are well placed to lead AI education through critical appraisal skills, evidence expertise, and established roles in governance‑aligned information practice. As AI technologies continue to evolve, librarian‑led education offers a sustainable model for supporting confident, informed, and responsible AI use across healthcare organisations.
Acknowledgements
We acknowledge Digital Health Services and the NSLHD AI Community of Practice for governance support and thank NSLHD clinicians whose questions and feedback shaped the program. We also acknowledge NSLHD Libraries staff for their ongoing commitment to building organisational capability in responsible AI use.