Problematizing the Public Domain: emerging issues for open access to cultural heritage against the backdrop of AI, machine learning and computational processing

Abstract

Advocates for open GLAM (Galleries, Libraries, Archives and Museums) have worked diligently over the past two decades to convince lawmakers, cultural heritage institutions, and their decision makers that public domain heritage should remain in the public domain once digitized. This has resulted in the widespread publication of cultural heritage data for unfettered reuse, such as digital surrogates, collections data, transcriptions, metadata, and paradata. More than 1,600 cultural institutions and organizations around the world publish materials under open licenses and tools, resulting in more than 95 million digital assets and collections data being available for unfettered reuse. More recently, technology advocates and data and computer scientists have worked diligently to expose the harms and biases evident in LLMs and artificial intelligence technologies. What has received less attention is how these two fields intersect and why the should increasingly influence each other.

Using empirical data from the open GLAM survey (http://bit.ly/OpenGLAMsurvey), this paper maps the potentials of digitization and digital collections for new technologies that incorporate artificial intelligence, machine learning, and computational processing. It explores emerging trends and challenges evident in the global landscape of open GLAM activity, as well as the harms and biases surfacing in projects due to wider open access and heritage data availability. Rather than presenting the public domain as a neutral good, the paper demonstrates why and how we must problematize the public domain to support more ethical approaches to collections management, digitization, data publication, open GLAM participation, and wider reuse.

Keywords: artificial intelligence, public domain, copyright, open access, open GLAM, cultural heritage collections

JEL Classification: K11, O3, Z1