15 November 2024
Leah Maughan, Scholarly Communications Librarian, Northumbria University, Susan Mair, Copyright Advisor, Northumbria University
‘Oh my god, what the hell is this?!’ If we’re honest, this is a fairly standard cry from librarians. In this case, however, it was specifically aimed at AI—generative AI in particular. To the uninitiated, or even to those with some knowledge like us, the language and landscape of AI evolve so rapidly it feels like a losing battle to keep up. Do we really understand the terminology or even the purpose of some of these tools? And how on earth are we supposed to teach others about them?
We watched as institutions, including our own, started grappling with what AI might mean for undergraduate teaching and assessment, leading to new strategies and policies. However, it became clear that AI’s use in research wasn’t receiving the same scrutiny. How could we, as librarians supporting researchers, track the tools being used across multiple disciplines? How could we know what guidance to provide when researchers use AI tools? And how could we possibly keep up with it all?
The answer, of course, is that we can’t. But do we need to? We’re not life scientists, sociologists, or IT professionals. We are librarians, and in this AI-driven age, we bring well-established knowledge and skill sets to a new generation of tools.
Every Problem is a Nail…
With this in mind, we set off for the Academic Libraries North Conference 2024. While others have, quite justifiably, debated the appropriateness of applying AI to various aspects of research, we titled our workshop on research support and AI ‘Every Problem is a Nail, and I’ve Got this Hammer.’ Participants examined library research support, assuming that AI is being applied to every conceivable task. They considered research data management, publishing, and literature searching—reflecting on how a researcher might use AI for these tasks, and which library-related skills are essential to mitigate risks.
The skills required—evaluating sources, critical thinking, and copyright awareness—are long-standing library proficiencies. Our challenge is applying them confidently in the AI arena. Judging by other sessions at ALN24, many librarians are already doing this in a variety of ways. These excellent sessions are now available through the ALN website for anyone keen to learn more.
Making Friends…
To prepare, we conducted outreach with researchers and students at Northumbria, gauging their awareness of AI. The responses were modest—some mentioned coding, writing, and translation—but AI was often described as a ‘good friend’ or sounding board. The non-judgmental nature of AI tools emerged as a theme during the conference. We know that early-career researchers can feel stigmatised when admitting knowledge gaps. This might be a crucial point for librarians supporting this group.
The collected data was anonymised prohibiting us from discovering more from the person who submitted the statement that AI is a “friend” to them.
Current studies suggest the utilisation of AI as a substitute for human connection is prevalent in higher education. Mainstream media has covered this phenomenon, reporting widely on published academic research.
…While this behaviour does not directly concern library skills, it might be something to work on with our student wellbeing teams. Opportunities for collaborative research can be beneficial creating a more joined up service for students.
Navigating AI: Bias and Accuracy
From discussions with students and staff, other uses of AI emerged: literature discovery, interpretation, and identifying trends. However, serious concerns surfaced, including AI’s tendency to fabricate non-existent literature, plagiarise, and reproduce biased information based on its training data. For example, a study on medical research presentation via ChatGPT found that 47% of papers referenced were fabricated, with another 46% inaccurate. This raises important questions about how researchers use AI and how their research might be misrepresented.
Librarians possess transferable skills to guide researchers in avoiding such pitfalls. Our expertise in understanding citations, authorship principles, publication practices, and decolonising literature can be crucial for users verifying AI-generated content.
Large Language model’s operating principles can cause unintentional plagiarism, rather like an overly opinionated friend, the tool does not ‘know’ where its information comes from, only that it has ingested it at some point. It may therefore present a user with uncited work, or unintentionally biased information. For instance, when prompting ChatGPT to provide me with a publication strategy, I am offered a selection of potential (real) journals, but with a US/Eurocentric focus. Users may find that wider perspectives are not reflected in the answers they receive, especially those researching specific geographies and demographics.
Librarians have transferable skills to help guide and enhance their users experience in these tasks. Understanding research citations, authorship principles and publication practices. Knowledge gained through decolonialising literature and spotting predatory publishers could be adapted to help users verify information and understand its providence.
Crossing the Streams
AI can lead researchers in unexpected directions—unlike other systems, it doesn’t work within siloed content, AI can take you anywhere and if you don’t take control, it will take control for you. At ALN24, discussions moved across topics like EDI issues, wellbeing, copyright, ethics, and technical expertise. At Northumbria, we’re working across teams to build a more holistic support model for researchers using AI, partnering with Academic Technical Services and Research Innovation Services to help researchers make informed decisions about how and when to use it.
There’s no single way to handle AI; it will require continuous adaptation. Library professionals in HE are well-prepared to integrate AI into their traditional skills portfolio. We can frame AI as an enabling component in our services, helping researchers enhance their skills to navigate the information landscape of the future.
Librarians should be confident to promote our expertise and share our experience and opinions across our institutions, working collaboratively to deconstruct and resolve anxieties around the use of AI.
Back to the newsletter