Centering Humans in AI-enabled Manuscript Transcription

Date

2024-05-21

Authors

Brumfield, Ben

Journal Title

Journal ISSN

Volume Title

Publisher

Texas Digital Library

Abstract

How can artificial intelligence be used to support humans instead of replacing them? This presentation demonstrates how we integrated Handwritten Text Recognition (HTR) into the FromThePage crowdsourcing platform while keeping humans central to the transcription process.

We asked ourselves how can we introduce enough friction into our user interface that volunteers have to think rather than blindly trusting machine generated data, while still using AI to speed up their work and increase their pleasure? Our focus lay on enhancing user experience rather than on character error rates or model training.

We will discuss design considerations, implementation specifics, and the initial outcomes of user testing conducted by the State Library of New South Wales and their Digital Volunteer working group.

Description

TCDL 2024 Session 1B, Tuesday, 5/21/2024, 3:45 pm to 4:45 pm | Moderated by Adrian Shapiro, Texas Woman's University | Presentation | Software Development / Technical

Citation