Applying AI to digital archives: trust, collaboration and shared professional ethics

Abstract

Abstract Policy makers produce digital records on a daily basis. A selection of records is then preserved in archival repositories. However, getting access to these archival materials is extremely complicated for many reasons—including data protection, sensitivity, national security, and copyright. Artificial Intelligence (AI) can be applied to archives to make them more accessible, but it is still at an experimental stage. While skills gaps contribute to keeping archives ‘dark’, it is also essential to examine issues of mistrust and miscommunication. This article argues that although civil servants, archivists, and academics have similar professional principles articulated through professional codes of ethics, these are not often communicated to each other. This lack of communication leads to feelings of mistrust between stakeholders. Mistrust of technology also contributes to the barriers to effective implementation of AI tools. Therefore, we propose that surfacing the shared professional ethics between stakeholders can contribute to deeper collaborations between humans. In turn, these collaborations can lead to the building of trust in AI systems and tools. The research is informed by semi-structured interviews with thirty government professionals, archivists, historians, digital humanists, and computer scientists. Previous research has largely focused on preservation of digital records, rather than access to these records, and on archivists rather than records creators such as government professionals. This article is the first to examine the application of AI to digital archives as an issue that requires trust and collaboration across the entire archival circle (from record creators to archivists, and from archivists to users).