Promoting Open Science in times of AI: Do we grasp the interplay?

This presentation, delivered by Marie Alavi at the OPEN SCIENCE RETREAT hosted by ZBW on March 11, 2025, offers an awareness-raising examination of the relationship between Open Science (OS) and Large Language Models (LLM). The authors emphasise the significance of this analysis for a responsible deployment of LLMs in research.

The presentation highlights the challenges posed by the distinct logics between LLMs and OS - the former generating outputs based on mechanisms such as probabilistic modelling of token sequences, while the latter emphasises guiding thoughts of Research Integrity (RI) and Research Ethics (RE) as prerequisites of responsible conduct and dissemination of research. 

The authors conclude that promoting OS as widely as possible could help resolve this dilemma, thereby refining the quality of freely accessible knowledge. They propose connecting LLMs with scientific knowledge graphs as a potential solution for ensuring efficient and reliable usage of LLMs in research.