Archives and libraries are increasingly investing in mass-digitization, but until recently transcriptions of manuscripts were costly to generate. Handwritten Text Recognition (HTR) technology is now transforming access to our written past, producing increasingly accurate transcriptions for use by individuals and institutions, or providing material for further analysis. However, there has been little consideration of how this will affect archival and historical method. Considering the Transkribus platform as a community of practice, this chapter will report on a survey of users of HTR, undertaken as a reception study regarding the practical, methodological, theoretical, and ethical issues raised when inviting machine learning into historical archives. Evidencing Transkribus use by a diverse community, it is suggested that the scale and scope of transcriptions generated by HTR will require new approaches to both history and public engagement, while providing recommendations on how to best support the community applying HTR to cultural heritage materials.
|Title of host publication||Archives, Access and AI|
|Subtitle of host publication||Working with Born-Digital and Digitised Archival Collections|
|Place of Publication||Berlin|
|Publication status||Accepted/In press - 15 Mar 2021|
|Name||Digital Humanities Research|