Diversity, Sustainability, and AI: Key Takeaways from RLUK's Virtual Conference on Open Science

RLUK’s annual conference, which is held virtually in early spring, has been bringing together thought leaders from the UK-, Irish-, and international library community since 2013. The current chair of the RLUK, a consortium comprising 39 of the most significant research libraries in the UK, is Masud Khokhar, who not only steered some sessions skilfully but gave strong insights and food for thought throughout open discussions, such as on the topic of diversity and equity and community engagement; picking out one of his thoughts from the session on the work libraries undertake in support of the UNs various sustainability goals, Masud reminded us of the growing carbon footprint of digital collections and infrastructure, in which all stakeholders need to scrutinize their own climate impact vis-à-vis SDG 13. He picked out commercial publishers, who should be more proactive in pursuing action for the health of the climate beyond hollow webpage promises. On a similar note, Brett Waytuck of the University of Regina mentioned that more analysis is needed to ascertain the climate impact of cloud libraries versus the benefits of open, shared services. My takeaway from the discussion around the sustainability of Open Research is that a willingness for more positive change is palpable, but that, as so often, the picture is grey and complex.

The keynotes were informative and enjoyable and, in fact, hit very stark notes, both in raising hopes that we are on the right track to improving the hygiene of science communication through the open science movement, as well as laying bare the downsides. The strongest statements were made by Alison Mudditt, current CEO of the Public Library of Science, who noted that her primary concern is that this high-minded transition to open access has shifted from being this movement for systemic change to becoming more of a business model for supporting the status quo. The statement was sobering, and, at the very least, pointed to the amount of work that is still required in phasing out the publish-or-perish model, the overreliance on traditional research evaluation metrics, and an overall sluggishness and resilience to change within an environment of profitability of existing models (who wants to cut the hand that feeds them). Alison ended on a brighter note when highlighting some of PLOS’ initiatives launched to foster the much hoped for systemic shift, which is just about underway.

As a fellow German national, I am always curious to learn about other Germans who are trying their luck outside of the shelter of German-speaking countries, and I was delighted to observe the keynote of Torsten Reimer, originally from Munich, now librarian and dean at the University Library of the University of Chicago, who cited the alarming recent trends of increased censorship within the American education system, and the initiatives that the University of Chicago Library is undertaking to confront, counter, perhaps reverse, these, thus standing up for freedom of speech and freedom of expression; namely, the institution’s building of a comprehensive collection of around 1,500 banned books – mostly books challenged due to being written by or about people of color or members of the LGBTQIA+ community – which is made available throughout digitally throughout Illinois and the wider US, putting itself in a vulnerable position, taking a clear stand, which I greatly admire.

Naturally, AI was still a dominant topic at the RLUK event. Torsten Reimer pointed out that it is an issue for libraries using generative AI capable of proliferating inaccurate data, as it is the very mission of every library to disseminate accurate, trusted data. Kate Robson Brown of University College Dublin cited the recent endeavors by governments in the UK, where a framework of AI safety principles was developed in Bletchley Park, as well as the EU AI Act, as positive trends for monitoring a largely under-regulated field; however, she also voiced surprise at the exclusion of the voices of librarians from inputting into these developments – even though librarians are the experts in the accessing and dissemination of knowledge. Andrew Pace, Executive Director of the Association of Research Libraries, cautioned for responsible use of AI in the face of plundering cooperations, another memorable noun phrase that still rings in my head.     

I loved the theoretical approaches being discussed but gravitated to the examples of open science models in practice. Natasha McCarthy, Associate Director at the Royal Academy of Engineering, highlighted projects in which AI makes a real change to science - the acceleration of protein structure predictions in the AlphaFold DB (developed by Google’s DeepMind), among others.

In a separate talk on AI, Andrew Horne of the University of Edinburgh reported on the power of LLMs in automating literature reviews and eradicating vast, unintelligent work. I was fascinated to learn that the pilot application of a combination of LLMs and chat interface within the study field of microfluidics reduced the overall literature review time by 90%! If automation at such a scale can be achieved across subjects, then up to 50% of funding money currently flowing into (or is wasted on) manual reviews can be saved, and more barriers to research are taken down, a fantastic unleashing of the power of AI tools.

The current status of open science and the various directions in which it can – or may – evolve were debated in several fascinating Roundtable discussions. Diamond Open Access, for example, a community-led infrastructure, is pondered by many and pioneered by some, and the University of Cambridge Library, for example, reported on implementing a Diamond OA pilot for journals using DSpace. While Diamond OA is a model that will work for some, due to the investment cost and sizeable technical support needs, it is clearly not a system that will work for all.

The efficacy of transformative agreements in the UK was also up for debate, a timely discussion given JISC’s recent review, which showed that the UK came out ahead of the global picture, transitioning to open access faster than other countries by a small margin and achieving high levels of funder compliance. The overwhelming sense is that TAs are not achieving transformation widely and quickly enough but have led to substantial savings in the sector and justify a confidence in negotiations with commercial publishers, and a desire to appear be more bullish to address red lines in the evaluation of current transformative agreements.

I will end with Alison Mudditt, who posed a few succinct questions to librarians at the end of her keynote. If transformative agreements are not delivering what they should, then what is next, she asked. At the current flip rate, it would take big publishers at least seventy years (!) to flip their portfolios fully towards open access. Apart from bullishness, and new models that leverage greater collaboration, diversity, and new approaches, change should come along at a more radical speed. 

I have never attended a fully online conference, so it was all a bit of a novelty to me. I found that the format of presentations, followed by Q&As, interspersed with Roundtables and very powerful keynotes, worked supremely well. On the third day, four interactive workshops (two of which were explicitly around AI-related topics) rounded off the RLUK conference. I found all elements of the conference inspiring and a lovely way to check in with a lively and passionate community of mostly librarians, who are skilfully taking the temperature of the debate of the direction of open science, checking in on the patient, putting forward ideas and solutions to some of the sector’s greatest challenges.   

Don’t take my word for it; instead, replay the sessions from the event on RLUK’s YouTube channel and get inspired yourself: https://www.youtube.com/@ResearchLibrariesUK.