AI and Alt Text for Archival Image Collections

Theresa Berger (she/her) is Digital Library Services Librarian at the University of Minnesota, where she oversees digitization and digital collections development for Archives and Special Collections. This presentation will discuss ways in which a closed ChatGPT model can be used to generate alt text (alternative text) in archival image collections, using a "human-in-the-loop" approach. Theresa will highlight both strengths and weaknesses of the tool, as well as suggestions for overall workflows, showing ways in which AI can be used to shift (not replace) resources in an effort to better serve our users, stakeholders, and our communities.