Invited Talk: Friend or foe

Metadata related to the conduct and communication of research, located in systems such as bibliographic databases, software archives, data repositories, and current research information systems, forms an important part of the research ecosystem. It is used in decision making around strategic priorities, distribution of resources, and evaluation of researchers and institutions. It is also used in assessment of the effect of policies (e.g. on open access and open data) and is instrumental in discovery, enabling researchers and societal stakeholders to fund and assess research results. Open metadata (free to use and free from restrictions on reuse) promotes transparency in assessment, equitable decision making and inclusivity in discovery, fully in line with open science principles. Developments in artificial intelligence pose particular challenges for open research information. One one hand, AI approaches can and do facilitate the extraction and processing of metadata from full-text articles and other research outputs, enabling more efficient workflows and potentially increasing the availability and accuracy of metadata in publisher systems, institutional systems (catalogues, institutional repositories, CRIS systems) as well as in metadata aggregators and bibliographic databases. On the other hand, there are important questions around provenance, transparency and quality control that need to be taken into account, as well as questions around data ownership and control and digital sovereignty. AI developments, especially around generative AI and training of LLM, also contributes to enclosure of metadata, with publishers less willing to make metadata openly available they can also license to AI companies to train LLMs, and even going as far as to have open metadata removed from bibliographic databases. More in general, there is a discussion in the research community about whether open availability of research outputs should include the use by commercial companies for generative AI applications, or whether such use should be restricted. This discussion extends to the availability and use of open metadata. This presentation will provide an overview of these developments and viewpoints as a starting point for discussion about the implications of (generative) AI for the creation and use of open metadata.