“Lost in Transcription?”: A case study of using AI and humans-in-the loop to create captions and transcripts for a library symposium

Abstract

In many regions, academic libraries are increasingly expected to generate recorded content as part of their services and operations, such as hosting public or academic events. These recordings require quality captions and transcripts to address the needs of users with disabilities as required by human rights legislation. Unfortunately, auto-generated captions cannot be relied upon to provide error-free outputs. This paper presents a case study of how we used AI and involved 35 volunteers from across North America to create captions and transcripts for recordings of a 3-day virtual accessibility symposium. This initiative was put in place to ensure all attendees had equitable access to the recordings. Our experience asserts the necessity of including humans in digital environments that integrate AI workflows.

Keywords: accessibility, captions, transcripts, artificial intelligence, collaborative initiative