New Voices RGU Student Series 2026 – Victoria Barsalou

Category: RGU Student Series 2026

In the 2026 New Voices Student Series, the CILIPS Students & New Professionals Community will be sharing the views of Robert Gordon University students from the MSc in Information and Library Studies.

With special thanks to Dr Konstantina Martzoukou, Teaching Excellence Fellow and Associate Professor, for organising these thought-provoking contributions.

Victoria Barsalou comes from France and Switzerland and holds a Bachelor’s degree in Ancient History and Religious Studies, as well as a Master’s degree in Ancient History from King’s College London. She lives in Bonn, Germany and is completing the online MSc in Library and Information Science at Robert Gordon University, wanting to pursue a career in academic or school libraries.

Teaching GenAI Literacy in the School Library

Artificial intelligence is one of the biggest scientific breakthroughs of recent years. Specifically, Generative AI (GenAI) has reshaped how we learn, search and study. While some have adopted these tools instantly, others worry about risks, misinformation or privacy. But in the school-setting, one fact is unavoidable: students are using AI every day, often without understanding accuracy, bias, hallucination, or privacy concerns. Research like Wong & Chiu (2025) shows that many students turn to GenAI as a primary information source.

The Mechanism of Misinformation

To design effective training, we must first understand how these tools work. Large language models (LLMs) like ChatGPT generate answers by predicting the next token in a sentence based on probabilistic models trained on sources written by humans. This raises questions about bias, data piracy, and reliability of output since the training data written includes copyrighted, biased content and false or dangerous information.

False information or fake citations are what researchers call “hallucinations”. Why this occurs is poorly understood; LLMs constitute what is referred to as a “Black Box”, meaning the exact train of thought leading to the output they are producing is opaque even to their programmers.

Librarians report that many students cannot recognise hallucinated citations or fabricated statistics (Foot 2024; Seales 2024). Similar findings appear in Oddone et al. (2024), who describe students’ difficulty distinguishing valid sources from AI-generated misinformation. What is needed here is awareness of the fact that just because LLMs can sound confident in their replies, their reasoning is murky and their output potentially misleading.

Why the School Library?

School librarians already teach media literacy, research and citation skills, and GenAI is right at the centre of these competencies. Frameworks like the EU/OECD AI Literacy (2025) urge educators to teach students the importance of not simply using LLMs as a tool but to reflect on the veracity of and the methodology behind their outputs.

Designing the Training: Domain Matters

Newer domain-specific AI tools have been demonstrated to hallucinate less than general-domain LLMs (Gu et al., 2021). OpenEvidence in medicine, for instance, draws its answers from peer-reviewed literature and produces inline citations that make claims immediately verifiable. This was empirically validated by Low et al. (2025).

Oddone et al. (2024) investigated this idea further and found that students should use different AI tools depending on what stage of their work they are in, from research to discussing plagiarism and supporting the writing process. It is increasingly clear that guiding students towards using the right GenAI tool will be an important part of a teacher librarian’s teaching. Poremba (2025) describes simple activities where students test AI claims against academic databases to learn to detect fake citations. These short exercises help students understand what AI can do and exactly where it fails, and to discriminate between different LLMs as they see fit.

A Framework for Student Guidance

Based on this approach, a practical training resource for librarians should focus on a small set of essential skills:

1. Teaching the basic principles behind LLM architecture and its associated pitfalls

• • 1) Hallucinations
  • 2) Domain specificity

2. Tool Selection

• • 1) Different use-cases matched to different models
  • 2) How to validate output

3. Ethical use of GenAI

• • 1) Tools as writing assistants rather than instances of plagiarism

We must encourage students to ask simple but important questions:
– Where does this AI get its information?
– Does this tool cite real sources?
– Is this AI designed for my subject?

One way to implement this would be short, embedded “micro-lessons” in which librarians present examples of AI-related pitfalls, such as hallucinated citations, comparing GenAI output with a database result, or choosing the right AI-model for their respective use-cases. Harrison et al. (2025) demonstrated that such activities could be integrated without requiring the context of specific units. In order to incorporate AI literacy into different, specific areas of the students’ curricula, librarians and teachers should collaborate.

Together, these skills give librarians a clear starting point for any GenAI training resource.

Conclusion

When students are taught about the aforementioned aspects of GenAI, they are more likely to produce better and more ethical work. Any training resource needs to start from the basic principles of GenAI in order for teacher librarians to contextualise different questions that students may have and make specific recommendations. GenAI now shapes how teenagers search, learn and produce work. With the right training resource, librarians can help students become safe, critical and responsible users of AI, a skill they will need far beyond the classroom.

Additional resources are available in the subject bibliography, which can be accessed by clicking here.

Reference list

ChatGPT (2026) School librarian teaching GenAI literacy to a high school class while conducting a history research project [Digital art]. Created for Victoria Barsalou, 22nd February.

European Commission and OECD (2025) Empowering learners for the age of AI: An AI literacy framework for primary and secondary education. Review draft (May 2025). Available at: https://ailiteracyframework.org/wp-content/uploads/2025/05/AILitFramework_ReviewDraft.pdf

Foote, C. (2024) ‘AI and School Libraries.’, Computers in Libraries, 44(5), pp. 32-34. Available at: https://research-ebsco-com.ezproxy.rgu.ac.uk/c/hin62q/viewer/pdf/cla3o3bydf?route=details

Google Gemini (2026), Student using GenAI tools for a research project [Digital art]. Created for Victoria Barsalou, 22nd February.

Harrison, A. (2025) ‘Inquiry, AI, and Pedagogy in the Library Classroom’, Virginia Libraries, 69(1), pp. 1-8. Available at: DOI: 10.21061/valib.v69i1.691

Low, Y. S. et al. (2025) ‘Answering real-world clinical questions using large language model, retrieval-augmented generation, and agentic systems. ’, Digital health, 11. Available at doi:10.1177/20552076251348850

Oddone, K., Garrison, K. and Gagen-Spriggs, K. (2024) ‘Navigating Generative AI: The Teacher Librarian’s Role in Cultivating Ethical and Critical Practices’, Journal of the Australian Library and Information Association, 73(1), pp. 3-26. Available at: doi:10.1080/24750158.2023.2289093

Poremba, M. (2025) ‘The Proof is in the Process: Incorporating Generative AI into Student Research Projects’, Canadian School Libraries Journal, 9(2). Available at: https://journal.canadianschoollibraries.ca/the-proof-is-in-the-process-incorporating-generative-ai-into-student-research-projects/?utm_source=chatgpt.com

Seales, D. L. I. (2024) ‘Adventures with AI in the School Library.’, Computers in Libraries, 44(7), pp. 30-34. Available at: https://research-ebsco-com.ezproxy.rgu.ac.uk/c/hin62q/viewer/pdf/tbcflurunn?route=details

Wong, E.K.C. and Chiu, D.K.W. (2025) ‘AI literacy instruction program in international school libraries: A qualitative study under the lens of the Big Six Information Literacy model’, Journal of Librarianship and Information Science, 0(0). Available at: doi:10.1177/09610006251331977

Gu, Y. et al. (2020) ‘Domain-Specific Language Model Pretraining for Biomedical Natural Language Processing’, ACM Transactions on Computing for Healthcare (HEALTH), 3(1), pp. 1-23. Available at: https://doi.org/10.1145/3458754