Kortext and HEPI blog image showing ChatGPT

Artificial intelligence, publishers and the translation of research

A reflection on making research accessible outside of academia

  • This blog on accessible research was kindly contributed to the Higher Education Policy Institute (HEPI) by Philip Carpenter, Pro-Chancellor at the University of York and an Advisory Board Director of Kortext, and originally published on the HEPI blog.
  • Philip offers some reflections on the recent HEPI and Taylor & Francis Policy Note, Why open access is not enough: Spreading the benefits of research.
  • HEPI and Taylor & Francis are hosting a free webinar on the issues in the Policy Note next week, on Tuesday, 31 January 2023 – sign up here.

Driving once between the Oxford and Chichester offices of the publisher John Wiley, I was passed by a plumber’s white van, emblazoned on the side with the words ‘John Abell, Solution Solutions’. I chuckled to myself about the inanity of business-speak, without stopping to reflect that I was just as guilty of using an inflated descriptor of my own work.

At the time, I was responsible for Wiley’s journal publishing business, which we called, rather portentously, ‘Research Communications’. But the descriptor made claims that were not entirely justified, for, generally speaking, publishers communicate research in just one format and just one way. They publish journal articles, whose format has not greatly changed in centuries. True, there are some efforts to produce lay summaries and the like, but that is not the norm.

Contrast this with a more recent experience when, as a member of the governing body of the University of York, I met with a number of the University’s research leaders. When I asked them about how the went about the business of ‘research communication’, I was taken aback to discover quite how active researchers are, and in how may different formats. Besides articles, they generate lay summaries, infographics, posters, videos and posts on social media, particularly now TikTok.

So there seems to be a gap between what publishers do by way of ‘Research Communication’ and what researchers themselves are willing to do to ensure that their work has impact – on society, on industry, on policymakers Maybe that is because publishers have been so preoccupied in recent years in the (largely successful) transition to the Open Access model. Maybe it is partly because there is no obvious commercial return for the publisher in engaging in various means of ‘translating’ research articles to maximise their reach and impact. Libraries pay subscriptions for access to articles, and authors / funders pay APCs for publishing services. There is not usually additional funding for wider forms of dissemination.

AI and Natural Language Processing may offer a solution to this impasse. Despite the current moral panic about ChatGPT and how students might potentially use it to write essays or draft exam question answers, there is a potential value in this and comparable systems in ‘translating’ the typically inaccessible language of the journal article into something that pretty much anyone can understand, including researchers working in other disciplines. This emerging technology offers a means to support the translational capability called for in the recent HEPI and Taylor & Francis Policy Note Why open access is not enough: Spreading the benefits of research.

Consider for a moment the current capabilities of the technology study platforms students in the UK typically use. As an example, the Kortext Arcturus platform (for disclosure, I am an Adviser to Kortext) which is used in well over 100 universities, can take a passage or chapter of a textbook, translate it on the fly into Chinese or read it aloud, or indeed read it aloud in Chinese or most other languages.

Conceptually, it is not a big jump from there to, for example, a surgical patient visiting a publisher’s website, downloading a research article and having it ‘translated’ on the fly to a version that a non-specialist can easily understand. Technologically, it is not a big jump either. These systems are already being developed and exist in beta form.

For non-academic readers, the benefits are obvious. For researchers, such a system would, with no effort on their part, enable their reach into the user community and raise the impact of their work. For publishers, this technology will likewise deliver usage and impact at minimal additional cost and do something to substantiate their claim genuinely to be in the business of ‘Research Communication’.

What’s needed for this to happen is for the technologists to continue to develop the systems and for the publishers to invest in and install them in their platforms. It also requires the wider community to take a more nuanced view of ChatGPT and the like: to explore their considerable potential contribution to research dissemination, not be blocked by worries about the potential corruption of teaching and learning in higher education.

In case you missed it: Discover the recent student research Kortext carried out in partnership with HEPI exploring the future of digital learning resources: student expectations vs. reality.

Leave a Reply

Your email address will not be published. Required fields are marked *

Name *