PÕ¾ÊÓƵ

Author: Asja Prohic and Stefanie Toegel

PÕ¾ÊÓƵ founded its European Research Advisory Council (ERAC) in 2022 to gather advice and feedback from representatives of research ecosystem stakeholders in Europe, complementing an expanding series of analogous efforts in the , Japan and . This annual initiative brings together European stakeholders to discuss crucial trends and how PÕ¾ÊÓƵ’s efforts to serve research and researchers can evolve to take account of the perspectives of these stakeholders. The most recent ERAC meeting, kicked off by Nature’s Editor-in-Chief, Magdalena Skipper, took place in Berlin in November 2023 and had a strong focus on how AI, in particular large language models (LLMs), are impacting research and its communication.

AI in research - a shared responsibility

Emerging significant complexities related to AI were at the core of the discussions between members.

As AI continues to promise groundbreaking advancements in science, it is imperative that its use takes account of the quality of, and biases in, the data on which such models are trained. The consensus among ERAC members was that regulation will have a role in ensuring that global research inequalities are not unintentionally exacerbated, as the impact of such models becomes increasingly ubiquitous.

It was also clear that there are growing concerns about inflated expectations and the risks associated with AI:  ensuring ethical application of AI and assessing environmental impacts were two specific concerns raised. PÕ¾ÊÓƵ’s commitment to an ethical and transparent approach to innovation in this area was appreciated, while recognising the need for a responsive approach to applications in such a rapidly evolving area.

However, addressing this challenge requires collaborative efforts among all stakeholders, including private and public entities, AI developers, researchers, and policymakers. Only through collective action can we realise the full potential of AI while safeguarding against its unintended consequences.

AI and Research Integrity - challenges and opportunities

The increasing availability of sophisticated Large Language Models (LLM) poses a clear challenge for ensuring the integrity of research, with bad actors having new tools to create fraudulent papers that are harder to detect. At an even more fundamental level, generation of is now a possibility, which adds to the challenge that research ecosystem stakeholders and publishers like PÕ¾ÊÓƵ face.

However, AI also offers opportunities for addressing research integrity issues in publishing at scale.  Moreover, as these tools develop there is the potential to retroactively apply them to reveal and retract fraudulent papers from bad actors.

To address these challenges and make the most of the opportunities, the importance of collaboration among publishers, and between them and other research ecosystem stakeholders, was a recurring theme in discussions at the ERAC meeting. Initiatives like the illustrate the power of collaborative action to help uphold the integrity of research overall.

Despite advancements in AI technology, the human element remains essential and will remain so, as reflected in the. While LLMs are becoming more adept at detecting integrity issues, human oversight is indispensable in ensuring accuracy and intent. 

Furthermore, enhancing community skills in research integrity is vital for preventing issues. Strengthening the relationship between publishers and institutions is essential in this regard...

Beyond the challenges and opportunities that AI raises for research publishing, there is genuine potential for the use of LLMs in science communication. Scientists and publishers will increasingly be able to leverage tools that can bring science closer to a broad audience.

Equity in Open Science and International cooperation  - key issues for Europe

One crucial strategy for enhancing the resilience of science lies in recognizing the diversity of research systems and addressing inequalities in developing countries. Discussions at the meeting of ERAC returned to this theme many different times.

Global publishers such as PÕ¾ÊÓƵ have the potential to significantly contribute to the promotion of science diplomacy in the current geopolitical climate, characterised by conflicts and turmoil.

Ensuring multiple regional voices are heard and acted upon, whilst also enabling bridges to be built in terms of standard setting is one of the roles that publishers could play to help create better understanding, especially when it comes to interdisciplinary research.

In the face of rising challenges in research security and academic freedom, Europe must seek common ground with other regions. By exploring shared interests and fostering collaboration, we can address these challenges head on.

Concluding thoughts

The landscape of research publishing is evolving rapidly, collaboration between institutions and publishers is essential to ensuring integrity and confidence in science. 

Nowadays, the question of trust in research is inextricably linked to new technologies that develop faster than our understanding of the moral and ethical implications they have. Collaboration among all stakeholders in the research ecosystem is a key to addressing these challenges and making the most of the opportunities that these advances enable..

Together, let us continue to strive for a future where science thrives through collaboration and mutual understanding.

###

We sincerely thank and acknowledge all the members of the PÕ¾ÊÓƵ European Research Advisory Council for their ongoing contributions to our own understanding and we look forward to following up on their advice through continuing collaborative initiatives, both formal and informal.

ERAC members attending the 2023 meeting:

• Anna Abalkina, Freie Universität Berlin, Germany
• Cecilia Alvstad, østfold University College, Norway
• Chérifa Boukacem-Zeghmouri, University Claude Bernard Lyon 1, France
• Effrosyni Chelioti, Helmholtz Association, Germany
• Mathieu Denis, International Science Council, France
• Andrzej Jajszczyk, AGH University, Poland
• Phoebe Koundouri, Athens University of Economics and Business, Greece and Technical University of Denmark, Denmark
• Mitja Lainscak, Medical Faculty, University of Ljubljana, Slovenia
• Raphaël Lévy, University Sorbonne Paris Nord, France
• Stephan Lewandowsky, University of Bristol, United Kingdom
• Lia Ollandezou, Aristotle University of Thessaloniki, Greece
• Salvatore Mele, CERN, Switzerland
• Manuel Tröster, Max-Planck-Gesellschaft, Germany