Build information literacy with AI: a teaching librarian approach

GenAI is rapidly changing the way we search for and produce information; digital and information literacy training must keep pace with this. University librarians often work at the coalface of such training, and so understanding how best to use generative artificial intelligence (GenAI) tools in this context is key.  

While the lack of depth and accuracy in AI-generated content has been widely criticised, GenAI can enhance aspects of the academic research process. And through critically assessing the validity of AI outputs, drawing on the depth of their subject knowledge and reflecting on the usefulness of AI in research, using it helps students to exercise and develop their critical thinking. 

Academic librarians must promote and demonstrate best practices to ensure that such tools enhance rather than replace human thought. 

Students’ use of GenAI tools, such as ChatGPT, to search for literature presents challenges that may not be immediately obvious. While some students argue that using ChatGPT helps them find papers related to their topic more quickly than searching through their university online library or an academic database, there are several limitations for students who exclusively search for their literature in this way. These include: 

• Hallucinated content: without critically evaluating the quality of the AI-generated results, students may undermine the academic integrity of their work
• A limited picture: AI tools such as ChatGPT can provide a response to a question or query but cannot tell a student which search terms they used, where they searched for the literature and any criteria they used to restrict search results. These are important parts of what students must learn to do as researchers.  

Ultimately, if we do not guide students through how to engage with AI tools in the search process they end up with patchy literature and risk losing and underdeveloping information skills. 

• Spotlight guide: Bringing GenAI into the university classroom
• Is critical thinking the answer to generative AI?
• Promoting ethical and responsible use of GenAI tools

We provide AI literacy training to build these skills, and to help students understand how to use AI to enhance, rather than replace, research processes. Our sessions involve:

1. A brief introduction to what AI and GenAI are and the differences between widely available tools and academic database tools, such as Scopus AI. We then ask students to evaluate the efficiencies of each tool based on their search task. For example, knowing that AI academic database tools produce peer-reviewed papers and that ChatGPT is more likely to provide hallucinated results helps them make informed choices.
2. Research question creation. Rather than a fully crafted research question, students come up with a “how”, “why” or “what” question containing key concepts or ideas of interest. We then encourage students to add their question to an AI academic database tool and review the response. At this stage, we introduce students to key concepts and functions of creating a search string (such as truncation, phrase search and Boolean connectors). 

Next, drawing on their knowledge of search structures, we ask students to 

• Review the keywords and phrases used to search
• Consider how the AI tool might have rephrased the research question
• Analyse the generated summary it provides as an answer to the question. 

For example, they might reflect on where the tool is transparent about the search terms it uses and where it has accessed the papers. Or where the tool has not quite structured and formatted the search effectively.

These reflective questions help prompt discussions:

• Which search terms are useful to you?
• Are there search terms you would take out? Why?
• What search terms would you add? Why?
• What has AI forgotten?
• How effective has AI been in aiding our search process? How do you know this?

We then ask students to rewrite their search string and run this in the university library search, exploring the range of results that appear from the many subscribed databases. We ask students to apply filters and limiters (including date of publication, geography, information type), explore the information on offer and reflect on the differences in control, agency, transparency and clarity when comparing both AI and non-AI searches.

Students that engage in these sessions benefit from practical experience using a range of AI tools without judgement. 

When used as part of the pre-searching phase, students can be inspired by the ideas and responses from such tools, but remain in control of decision-making by drawing on their understanding of how to search and what to search for. 

Callum Perry is an academic and teaching partner at the University of East Anglia’s Faculty of Social Sciences.

If you would like advice and insight from academics and university staff delivered direct to your inbox each week, sign up for the Campus newsletter.