As you know, AI technologies are advancing rapidly and have the potential to greatly affect how we handle, preserve, and make accessible archival materials. I am interested in hearing about your experiences, thoughts, and perspectives on this topic. 

- Have you started using AI tools for your archival processes, such as for metadata generation, digital preservation, or content analysis?

- What opportunities or challenges have you faced while using AI in your workflows?

- Have you identified any ethical considerations or best practices when applying AI in an archival context?

-Are there any uses for the technology that have made your life easier.

I'm especially interested in collaborating on projects or initiatives that delve deeper into these areas. Whether you're an AI enthusiast or just curious, I believe we can all benefit from sharing our insights and experiences.

Let's use this forum post as a space to exchange ideas, ask questions, and even develop collaborative efforts that advance our profession.

Looking forward to a stimulating discussion.

Best regards,

Roberto Nanes

Archivist

University of Houston - Clear Lake

We have used some auto-transcribing, mostly Youtube captions which is AI based but we have started to look into other options.

I have gone to a couple From the Page webinars. So, I am not an expert but I am curious about the ways in which this could be a tool.

As an archivist at UHCL, I've been exploring practical ways to integrate AI into our workflows since the release of ChatGPT. One challenge we often face is the time-consuming task of assigning subject terms to ensure our finding aids are accessible through library catalog systems and ArchivesSpace. While AI tools like ChatGPT have well-documented limitations in accuracy and data dependency, I've found them particularly helpful in this area.

Over the past few months, I've used ChatGPT to generate suggested subject headings. Its analytical abilities allow it to parse content and recommend relevant terms, effectively supplementing our efforts. Of course, I always review and verify these suggestions, but this approach has streamlined the process and could be a valuable tool for our team.

When I first began applying this, I would not include the terms that I had selected. Chatgpt would return results that were very similar to the terms I had selected, but its suggestions seemed to cover subjects and names I had overlooked.

--R. Nanes
UHCL Archives and Special Collections