Questions with Difficult Answers : AI technologies and Open Community Infrastructures
As the dramatic growth of AI technologies has forced every sector to face its own challenges, scholarly publishing has not been immune, and, as a community-based open infrastructure Coalition Publica has faced its own particular set of demands in recent months. From a steep rise in automated access to the Érudit platform, to users of PKP’s Open Journal Systems (OJS) grappling with understanding inconsistencies in the volume of articles downloads, to editors seeking guidance from scholarly communication experts on adopting AI policies for their journals, the constituent communities of Coalition Publica have been considering the impact of this growth.
Since textual data constitutes the prime matter for the training of large language models and machine learning, the world’s main AI developers have turned to harvesting textual documents en masse. Besides server overload, the distortion of the consultation and download statistics is of special concern since they are key indicators for libraries, journal editors and research communities. The impacts on the discovery of scholarly content are yet to be understood. This situation also raises ethical-legal issues regarding data governance, transparency of use, and copyright. The ways in which these resource intensive technologies can and ought to be included in the services provided by open, community-led infrastructures is also an uneasy question.
This panel brings together three perspectives on the issues raised by the increasing prevalence of artificial intelligence technologies, based on experiences of Coalition Publica and its community. The presentations will focus on the measures implemented by Érudit to address the massive harvesting of the platform, the impetus for exploring AI-related features in OJS, and the support offered by librarians in terms of editorial policies for journals. It will explore the ways in which open and public infrastructure navigate the tensions existing between AI, open access and services sustainability.