TCDL 2025

As artificial intelligence (AI) becomes more integrated into libraries, questions arise about how to transition AI-driven research and development projects into reliable, production-ready tools for both internal and external users. This panel will explore these challenges through two case studies, offering insights and practical next steps for those navigating the shift from research to implementation.

The Oklahoma State University Archives continue to implement facing recognition models for photograph retrieval. Combining AI technologies and annotation tools has significantly improved the discoverability of digital archival materials. Looking ahead, this work will help integrate several of the digital archival functions within a comprehensive workflow.

For several years, The University of North Texas Libraries and the Portal to Texas History have worked with a variety of tools to provide time-based captions for audio-video materials in our Digital Collections. This work started with several rounds of research funding and has now transitioned into standardized workflows for all AV material being added to the system.

This moderated panel will provide brief overviews of both projects, their histories, their progressions from research and development projects into production ready infrastructures in use at their institutions. Following the overview presentations, the moderator will present a series of general questions for each participant to answer with the goal of providing advice and lessons learned for the audience.