Participant Learning and Open Archives Through Artificial Intelligence and Citizen Humanities

This is the authors' abstract. We don't add key points for this paper.

Online archives host millions of scanned historical documents. However, images of old handwriting remain unsearchable and largely inaccessible without transcription. These limitations can now be mitigated by advances in Handwritten Text Recognition (HTR), which uses machine learning to convert images of handwriting into text data. In the HTR and Citizen Humanities project The Detective Section, the Swedish National Archives invited volunteers to build and refine a large, high-quality training dataset for an HTR application. This qualitative interview study explores how participants in The Detective Section perceived their involvement in the project in terms of learning and knowledge development. By applying a typology for understanding multifaceted learning and drawing on conceptual resources from science and technology studies (STS), the study highlights how participants perceived themselves as developing into epistemic subjects within an epistemic culture. We found that participants learn both individually and collectively, often in unexpected ways, through and beyond predefined goals and protocols. Their strong emotional investment and interest in local history leads to additional activities, which in turn result in more accurate transcriptions. We conclude by discussing refinements to participatory practices within the GLAM sector that can support volunteer learning and help democratize access to archival collections.