AI literacy for healthcare librarians – what does it mean for us?

What can knowledge and library services staff do about artificial intelligence (AI) literacy?

Artificial intelligence stands at the forefront of healthcare transformation. For healthcare librarians, understanding AI isn't just about keeping up with technology —it's about fundamentally reimagining how we serve our communities and improve patient care by accessing and analysing information more effectively.

Understanding the language of AI

Should I use AI?

  • Librarians should take a balanced approach to AI tools, promoting their use, but providing cautionary advice where appropriate.  The AI Safety Report 2025 outlines the benefits and risks. Check out this post on research integrity.
  • Yes, but follow national guidance and local policy and procedures if you can. Here are some sample policy statements of use with evidence synthesis:  Cochrane, NICE, RAISE & NHS AI & Digital Regulation.
  • Do not upload confidential or person identifiable information. Free tools often improve their data based on the content you enter, and what questions you ask.  Once you have typed something into an AI search engine you can’t take it out again! Make sure you are familiar with an AI tool before you use it.  Understand the GDPR implications and follow the guidance from NHS England Transformation Directorate.
  • Know the copyright implications. Think about who owns the information you are using or data mining.  Copyrighted material should not be uploaded to AI unless permission has been given. Consider the principles of copyright and generative AI and the NHS CLA Licence.   Look at the publisher’s website for terms and conditions e.g. creative licence permissions if the article is open source. Also check who owns the copyright of any AI-generated content under the terms and conditions, before re-use.
  • Human oversight is required. We are not yet at the stage of full automation of reliable evidence synthesis (Clark et. Al., 2025).
  • Be mindful of sustainability: there are large amounts of energy and water cooling requirements needed for AI processing (Scottish AI Alliance, 2025).
  • Be transparent: Under COPE Guidelines AI should not be cited as an author.  Example of how to cite use of AI is now included in most citation guides including the APA style, Harvard e.g. use of Claude with the prompt “Summarise what AI Literacy means for Healthcare Librarians with practical examples” in the creation of this article (see references for example).  Transparency also highlights librarian understanding of AI and supports raising the library profile.  You may wish to add a disclaimer to explain the process followed or invite the reader to use own judgement and check details prior to using the information e.g. “This evidence search (the output) was generated using AI. A librarian has reviewed and validated the content to ensure its accuracy (risk mitigation). We use AI tools to formulate the search strategy and summary based on publicly available abstracts (how it was used).  No material has been uploaded into the Large Language Model without copyrighted approval” (any additional message).

Practical AI applications in healthcare libraries

Literature search and systematic reviews

AI is already integrated into many search platforms to support semantic searching which understands context and intent rather than relying solely on keyword matching.  You will see it suggesting MeSH terms in the Knowledge & Library Hub.

For example, when a clinician requests information about "cardiac rehabilitation outcomes in elderly patients," AI can identify relevant studies that use different terminology like "cardiovascular fitness programs for geriatric populations" or "heart disease recovery in aged adults." This dramatically reduces the time required for comprehensive searches and when combined with natural language processing creates user-friendly interfaces. 

Knowledge Graphs can be created to create new links between different types of information and differentiate between similar concepts. It can be used in a predictive way to identify trends and identify gaps in the literature (Haque et. al., 2022).

There are several tools which can support different parts of the systematic review process. Khalil H, Ameen D, Zarnegar A. (2022) outline a range of validated tools for use at different stages.

Reference and consultation services

Alongside traditional evidence-based services, AI can help to create  knowledge maps to show how information is being used in an organisation (Boyes, 2022).

AI chatbots and virtual assistants have been tried in some services. These systems can handle routine enquiries about library hours, database access, and basic research questions, freeing librarians to focus on more complex enquiries which need human expertise and clinical context. (Yan, Zhao & Mazumdar, 2023).

Content curation and collection development

AI excels at analysing large datasets to identify trends and patterns that can help with collection development decisions. It has the potential to be used for automated cataloguing and workflow processes.  It can be used to allocate metadata, or improve catalogue searches with personalised content curation and predictive search (Jothimani, Anandraj, & Aravind, 2024).

There is the potential to use it to support current awareness bulletins, to support sourcing articles, and to create generative summaries relevant to your organisation’s focus areas.

Supporting researchers

Data Librarian is an emerging role which supports with the curation of data sets.

AI tools are available which can automatically classify and tag research datasets, making them more discoverable and reusable across different projects.  Tools are also being developed to support qualitative and quantitative data analysis.

There is a useful UK Research Integrity Office article on questions and challenges researchers should reflect on before they use AI.

The role of the librarian in AI literacy

Understanding AI’s capabilities and limitations forms the foundation of AI literacy. This includes recognizing when AI tools are appropriate for specific tasks, and when you still need a human being. AI excels at pattern recognition and data processing but struggles with nuanced clinical judgment and ethical considerations that require human oversight.

Evaluating AI-powered tools requires new assessment criteria beyond traditional software evaluation like Dijkstra, Greenhalgh, Mekki & Morley’s (2025) How to read a paper involving AI. Healthcare librarians must consider training data quality, algorithmic bias, and transparency in AI decision-making processes. Tools that cannot explain their reasoning or reveal their training methodologies may not meet the evidence-based standards expected in healthcare environments.

Data literacy becomes increasingly important as AI systems require clean, well-structured data to function effectively. Librarians can support critical evaluation of how data quality affects AI performance and learn how to prepare datasets for AI analysis, including formatting them properly, creating metadata and identifying bias.

There are also ethical challenges. It is important that we understand the systems and raise risks of breaches in copyright, privacy and confidentiality where appropriate.  We already have an awareness of bias and have a responsibility to raise concerns around systems that could perpetuate health disparities, or limit access to information for certain populations.  We can advocate for transparency in AI-assisted services to understand influences on decision making and be transparent in our own use of AI.

With the increase in AI influenced articles, fake news and deep fakes, we need to understand AI and use it to fight, not spread or create misinformation.

Preparing for an AI-enhanced future

The integration of AI into healthcare library services will accelerate over the coming years, making AI literacy an essential professional competency rather than an optional skill. Healthcare librarians who develop AI expertise now will be better placed to lead their institutions through this technological transformation.

Professional development opportunities in AI literacy are expanding rapidly, from online courses and webinars to conference sessions and certification programs.

Collaboration with ICT departments, clinical informatics teams, data scientists, and service providers creates opportunities for librarians to contribute their information expertise to AI projects at the same time as developing hands-on experience with the technology.

Learn more

CILIP (2025) CILIP AI Hub https://www.cilip.org.uk/page/AI

Cox, A (2021). The impact of AI, machine learning, automation and robotics on the information profession. https://www.cilip.org.uk/page/researchreport

FedIP (2025) The FedIP Hub https://www.fedip.org/fedip-hub

Future NHS (2025) AI Virtual Hub https://future.nhs.uk/AIVirtualHub

Intellectual Property Office (2024) Copyright and AI: Consultation https://www.gov.uk/government/consultations/copyright-and-artificial-intelligence/copyright-and-artificial-intelligence

NHS AI and Digital Regulations Service for Health and Social Care (n.d.). Understanding regulations of AI and digital technology in health and social care. [Accessed 27 May 2025]. Available from: https://www.digitalregulations.innovation.nhs.uk/

NHS England (2025). Current and Emerging Technology in Knowledge and Library Services. [Accessed 27 May 2025]. Available from:  https://future.nhs.uk/HEE_KLSEmergingTech/groupHome

NHS England (2024). Digital & Data Framework: NHS Knowledge & Library Services https://future.nhs.uk/HEE_KLSEmergingTech/view?objectID=236739813

NHS Transformation Directorate (2025). Artificial Intelligence. [Accessed 27 May 2025]. Available from: https://transform.england.nhs.uk/information-governance/guidance/artificial-intelligence/

NHS England Workforce, Training & Education (2025). NHS Knowledge and Library Hub. [Accessed 27 May 2025]. Available from:  https://library.nhs.uk/knowledgehub/

A new email list LIS-AI-Literacy has been created.  Alongside this there is a community of practice which meets regularly on either a Tuesday afternoon or Friday morning.  It operates using distributed leadership to implement the PAIR Framework (Acar, O.A., 2023) through the framework of journal clubs, shared practice and challenges. Email [email protected] to be added to the calendar invites.

References

Anthropic. (2025). Claude AI (Sonnet 4). [Generative AI]. [Accessed 27 May 2025]. Available from: https://claude.ai/public/artifacts/e99026a2-5488-43bb-a905-5aa6529b2ad5

NHS Wales e-Library for Health (2025). Artificial Intelligence (AI) for research, library and information. [Accessed 27 May 2025]. Available from: https://elh.nhs.wales/support/artificial-intelligence-ai/

Ms Susan Smith