Structuring Content for Discoverability and Accessibility: AI’s Expanding Role in Academic Publishing

Cast your eye over the agendas of academic publishing conferences over the last couple of years and you see plethora sessions, panels and talks on the future of Artificial intelligence (AI) in academic publishing.  However, if you’re working in academic publishing, you may know that AI is no longer a distant concept; it is already transforming workflows, particularly in the production stage of the research lifecycle.

While the editorial and peer review phases have attracted much attention and debate on the appropriate applications of different technologies, production – often the quiet engine room of publishing – is where AI is already having a profound and immediate impact, without much discussion.

AI-driven technologies are available, accessible and currently being deployed in a variety of purposes from automating and enhance typesetting, copyediting, through to metadata enrichment.  At S4Carlisle we are using natural language processing tools to identify inconsistencies in language, suggest grammar and style corrections. Such tools are even able to flag potential ethical issues such as duplication of content (i.e., plagiarism checking), missing disclosures, or references made to retracted works.  Publishing suppliers, including here at S4Carlisle are increasingly using automated XML conversion tools to standardise article formats and references, ensuring rapid delivery across multiple platforms and repositories.  These efficiencies not only save time; they improve quality control and free up human editors to focus on higher-level review.

Structuring content for discoverability is another area where AI shines. Machine learning algorithms can extract and standardise keywords, assign article categories, and generate abstracts or summaries; all of which improve downstream discoverability and indexing. AI tools can also auto-generate alternative-text for images, ensuring accessibility compliance without adding to production timelines and through retro alt-text generation provides opportunity to open back archives to new users. AI also can help in maximising access to content in terms of understanding and applicability through create supplementary content such as lay summaries, generating PowerPoint slides, question banks etc. to enable real-world application of research.

And whilst this is the reality we’re living in; it’d be remiss not look to the horizon to where the potential extends further.  AI could enable real-time content enrichment – integrating datasets, linking to preprints or related research, and even generating lay summaries tailored to non-specialist audiences.  Intelligent tools might soon predict production bottlenecks, optimise scheduling, or support multilingual typesetting and translation. It’s in the S4Carlisle pipeline so watch this space!

Despite its promise, responsible deployment of AI remains essential and maybe there should be more conversation about how it is currently employed. What is indisputable at the moment, and what we can attest to at S4Carlisle in our adoptions, is that human oversight, transparency of tools, and alignment with research integrity principles must remain central to any adoption and application.

As academic publishing continues to evolve, as the AI use becomes prevalent, it’s clear for the near term, AI will not replace the role of skilled professionals. As we’re seeing in our adoptions and role outs, AI is empowering staff – making production more agile, accurate, and aligned with the needs of an open, global research ecosystem.

To find out more about how S4Carlisle can help you integrate AI tools into your workflows please get in touch with us https://www.s4carlisle.com/contact

Further reading:

• STM Trends 2029 – STM Association. A comprehensive future-focused report which includes current AI use cases in publishing.https://www.stm-assoc.org/what-we-do/strategic-areas/standards-technology/stm-trends-2029/

• ALPSP report on AI Tools in Scholarly Publishing (2023) Offers a snapshot of tools currently in use and publishers’ attitudes. https://publicationethics.org/topic-discussions/artificial-intelligence-understanding-current-guidance-and-tools (check “Artificial Intelligence Tools in Scholarly Publishing”)

• COPE discussion document on AI in research publication (2023) Focuses more on editorial ethics but addresses automation across workflows. https://publicationethics.org/guidance/guideline/ethics-toolkit-successful-editorial-office