ALPSP: AI = Friend or Foe for Protecting Research Integrity?

Feb 6, 2024 | Scholarly publishing

The ALPSP panel for this discussion on AI and the impact it is having within academic publishing was made up of Nicola Davies, IOP Publishing (Co-chair), Helene Stewart, Clarivate (Co-Chair), Meurig Gallagher, University of Birmingham (Speaker), Matt Hodgkinson, UKRIO (Speaker) and Jennifer Wright, CUP RI Manager (Speaker).

It was fascinating to see how the conversation around AI has moved on within a few months regarding this technological advancement. Institutions, publishers and journal stakeholders all have a concept of AI and are developing policies and guidance about how we should be using it and are underpinning the “what for?”.

Many of us by now will have tried asking a Large Language Model (LLM) to write a paragraph or create an image using Artificial Intelligence. It’s brilliant to watch how quickly tasks can be created, large blocks of text are generated at an immense speed, and right before us we see how quickly human intelligence can be mimicked.  This notion was highlighted in Meurig Gallagher’s presentation; that essentially AI is trying to act as human as possible using the instructions it has been provided with. However, when these tools are posed with mathematical equations it does not have the knowledge to apply the learning and therefore can spectacularly fail! These “gaps” therefore build into the guidance stakeholders need to be aware of when creating policy around AI – it cannot be relied upon solely to do the work. Matt Hodgkinson developed this further and shared many caveats that researchers and general users might come up against when using chatbots:

• Many LLMs are unvalidated for scholarly uses.
• References should be fact-checked as they can be falsified, therefore it is important to check sources and supporting literature.
• The quality of evidence is not assured.
• Outputs may be based on using out-of-date information based on “old” training material.

The ominous but noteworthy warning was circulated that “if you are not an expert, you will be fooled by fluent but incorrect outputs”.  Therefore, all of us involved in scholarly publishing need to be mindful of these contributions and check author statements within articles to assess whether an LLM has been used. Of course, one of the largest threats we are witnessing is the output of paper mills and their use of AI could lead to the tool’s collapse as its knowledge bank is infiltrated with “fake” data, which if left undetected will pollute the pool where the data is extracted from.

Nonetheless the principles of Research Integrity can be applied to the use of AI-generated content and Matt shared this slide to disclose how these principles are applied:

UKRIO presentation at ALPSP 2024

Dr Jennifer Wright from Cambridge University Press shared with the audience how to implement transparency which is really the crux of what many VEOs are looking at. It was suggested that AI declarations should be included within image captions, acknowledgements, and methodologies if applicable and the details that should be shared include the type of model that was used, eg: CHATGPT, and how and when it was accessed. It is also important to include any additional COI statements because of the use of the model. Looking forwards, Dr Wright elaborated on future considerations and posed some important questions around reporting standards: What will the impact be of AI on the scholarly record? How could/should/will research and publication practices change? How will concepts such as retractions be enforced? Can a bot retrain itself?

The challenges are still clearly evident with AI. However the more we progress and understand how it can be used, trust markers can be identified to validate the outputs. As long as scholars use and do not abuse the tech, we could watch something incredible unfold!

Unleashing the Power of Artificial Intelligence

Dec 2, 2023 | Scholarly publishing

AI is inevitably going to infiltrate all our lives in some way or another in the near future; learning how we shop, communicate, write, create and plan our lives. We therefore also need to look at adapting our ways of working to benefit from these technological advancements. Working alongside this adaptive new tech, and generating new guidance and principles, will enable us to harness and nurture it. We can create preventative methods to stop bad actors abusing and infiltrating the systems we have in place to educate and teach.

At the NEC Annual Publishing Conference (7 November 2023, London), the keynote was delivered by David Smith from the IET who looked “Back to the Future!”; highlighting the importance of an article published by Darcy DiNucci. Fragmented Futures (published in print, 53.4, 1999) demonstrates technological growth and that the Web she wrote about was only the beginning…how things have changed in 20 years! Essentially it is thought we are at a similar point with AI; it is new, raw and ready to be refined and developed.

Leslie Lansmann, Global Permissions Manager from Springer Nature, discussed how Large Language Models (LLM) such as ChatGPT are ingesting content, and this is not yet fully disclosed by AI companies. This is important to monitor as we must maintain the stewardship for the content and protect copyright and protected manuscripts. As much as AI is currently learning – it probes and reiterates content – it does not understand the deeper context behind the language. The publishing industry is however having to react to the developments in technologies, many publishers are imposing bans on AI content, and introducing new and different policies.

The discussion around authorship is constantly developing and debated – should research be done using AI? Can it help an author whose first language isn’t English produce a more succinct piece of work? If the data is accurate and the same research principles are adhered to, maybe we should move towards incorporating it into our practices. This notion was delivered by Anastasia Toynbee from the Royal Society of Chemistry who was looking at the problem of non-native English speakers and how these tools could help. The key feature of this was that a problem had been highlighted and AI was being used to support it – not the other way around.

It became clear from all the speakers how important it is to identify the problem initially and use AI tech/tools to help with it, rather than decide how to harness and squeeze new systems into processes that are working well. Ian Mulvany at BMJ really brought home this idea that we as an industry need to balance risk vs opportunity. AI has perception, however no intention to act; therefore we are in a position through governance, policy, and stewardship that we can lead AI to improve processes and not be reactive and in fear of the unknown! Andy Halliday , Product Manager at F100 iterated the benefits and pitfalls of AI and how humans can help harness this tech and enable it to support our ecosystem and develop a sense of AI preparedness.

We are in the awakening of AI. The box has been opened and we all have access to create new and exciting content, images and access information much easier than ever before. As the discussions continue it will be really exciting to see how developments are made, what fixes it can be used for, and how policy and guidance are updated to meet the demands of users.

Correcting the scholarly record and dispelling myths

Oct 16, 2023 | Scholarly publishing

Following the UKRIO workshop hosted by IOP Publishing and Karger on 20th September 2023, we discuss here the principles required for correcting academic literature and the key players responsible.

Post-publication correction notices are used to update or append research using neutral and factual terminology. Mistakes can be made, and post-publication corrections are not used to punish authors/journals. Corrections are not always a fault with the research, it could be an honest error.

Notices should follow industry standards and include key elements, such as: DOI, title, volume/issue number, year of publication and a description of the error and any actions taken to remedy the research.

The original article is not usually updated; however, it can be amended if it warrants legal or privacy concerns. This decision will be in accordance with the publisher’s policy and best practice. For example, a health journal may update drug doses if it would impinge upon patient care – this would be outlined in the notice and the content updated. The aim is to be transparent in the notice and include bi-directional linking. The notice should appear online and in print.

Types of Notice

1. Corrigendum. Usually an error introduced by an author.
2. Erratum. Usually an error introduced by the publisher.
3. Retraction. The most serious type of notice, following a full investigation.
4. Publisher’s note. Used to notify that an error may be in the content/under investigation.
5. Expression of concern. Advising the reader that there might be errors or untrustworthy content.

Withdrawal

Best practice is not to erase the content – a withdrawal notice, which is deemed the most serious type of correction, means that the DOI remains but the PDF is removed, not to cause detriment to the scholarly work.

Myths and Barriers to Correcting the Scholarly Record

Myths

• A correction does not always mean there is something ‘wrong’ with the research.
• A publisher’s responsibility for their content does not stop at the publication.
• An author doesn’t want to hear if you spot a potential error in their research.

Correcting the record

• Errors happen! Correcting the record needs destigmatising and normalising through education and transparent communication.
• Publishers must be willing to correct inaccuracies transparently with the support of all parties involved in the research ecosystem.
• Researchers should be willing to receive communications about their publications. Comments should be neutral and non-accusatory.

Standards are set by multiple bodies, including ICMJE, COPE, STM, and PubMed, which form a basis of recommended principles. Published content is a snapshot in time and should not be updated to reflect recent events/changes (for example, affiliation updates).

Who Decides What Needs to be Corrected?

This should be done in a partnership which can include the publisher, author, editor and editorial teams, depending on the query. For example, a plagiarism investigation will require more input from all involved as opposed to a typographical error in a name. Accuracy of publications must be maintained by all members within the ecosystem to uphold the scholarly record, which includes publishers, authors, readers, reviewers, editors and research institutions.

Publishers

• Need to have checks and balances in place to avoid inaccuracies being published.
• Correct inaccurate content in a thorough and timely manner using transparent language.
• Investigate concerns brought to the journal regarding the accuracy of content.

 Authors              

• Have a responsibility to avoid errors being introduced – thoroughly checking the content at pre-publication checks.
• Inform the publisher of any inaccuracies they identify in their own work.
• Inform co-authors of any inaccuracies discovered, whether accidental or intentional.
• Cooperate with investigations into concerns about accuracy of publications.

Readers

• Have a responsibility to report suspected errors in publications – this should be done neutrally to a body with responsibility for accuracy of the publication.

Reviewers

• Have a responsibility to review a manuscript critically and provide a succinct review. They should also report concerns with content to a appropriate body who has responsibility for accuracy regarding the publication.

Editors

• Have a responsibility to critically analyse manuscripts and report suspected errors.
• Investigate errors brought to their attention.
• Collaborate with the journal or publisher whilst an investigation is pending, bringing their subject expertise.

Research Institutions

• Have a responsibility to promote responsible research through education and foster a transparent research culture.
• Required to have a mechanism for reporting and investigating potential.
• Report the outcomes of the investigations to the publisher affected.

What is the Impact of Correcting Content?

It is important to correct and not remove content. Corrections will always be a customary part of maintaining the scholarly record and should only be done if necessary.  Removing or editing content could impact a researcher’s career. Retractions are the most serious type of notice that can be issued and can have a serious impact on the career of a researcher. Indexing services can be impacted, by splitting citations. Incorrect indexing can cause issues for journals, authors and publishers. Google Scholar scrolls every 6 months and therefore it does take time for services to be updated regarding notices such as retractions and withdrawals. Publishers must be responsible with post-publications to prevent inaccuracies in the scholarly record.