Artificial intelligence is often painted as a threat to culture and the arts. But museums, archives and libraries around the globe have been quick to recognize that AI — when used thoughtfully — can be an ally. 

No one wants DALL-E to usurp Dalí, but leveraging AI tools to expand access to collections and improve users’ experiences? That’s an exciting concept for librarians, archivists, curators and other stewards of cultural heritage.

“We didn’t run away from this technology,” says Abbey Potter, senior innovation specialist in the Digital Innovation Lab at the Library of Congress. “There’s so much promise. How could this benefit our organizations? How can this benefit our users?”

Potter earned her Master of Science In Information from the University of Michigan School of Information in 2005. Over the past two decades, she has worked at the intersection of digital innovation and cultural preservation. 

Now, as co-chair of the international community AI4LAM — shorthand for Artificial Intelligence for Libraries, Archives and Museums — she is collaborating with colleagues from cultural heritage institutions around the globe to share best practices for the responsible adoption of AI. This fall, she turned to her alma mater to help chart a path, enlisting MSI students to evaluate AI4LAM’s needs and make recommendations for its future. 

“Libraries, archives, museums, we are trusted organizations. We hold collections to preserve forever because they reflect something real and significant, and there are challenges to doing that, which AI could help with,” Potter says. “But these tools haven't been tested on historic material. They haven't been tested on culturally sensitive material. If accuracy is important — and libraries do have that higher standard of accuracy — we have to really spend a lot of time understanding how these models work.”

AI4LAM was established in 2018 as a knowledge-sharing community. Representatives of seven organizations — the National Library of Norway, Stanford University Libraries, Bibliothèque nationale de France, Ecole nationale des chartes, Smithsonian Institution, British Library and Library of Congress — make up the Secretariat, which coordinates events and supports the community’s needs. Membership in AI4LAM is individual and open to anyone who is interested.

AI4LAM has always been future-facing — its annual conference, held in a different country each year, is called Fantastic Futures. During monthly calls, members present on topics like using computer vision for automated tagging of archival images. They take on big questions: What are the business justifications for enhancing services with AI tools? What are the risks to organizations and staff? 

“The most interesting part of this effort right now is trying to get the machines to do what they can do really well,” Potter notes, “and then combining it with the deep expertise that library professionals have about the content.” 

But the organization’s own future is uncertain. For one thing, AI4LAM is not an organization, in the legal sense. This isn’t just an issue of semantics — becoming an organization would open up opportunities to apply for grants, hire dedicated staff and expand programming. 

This forward move also raises new questions: Would AI4LAM apply for 501(c)(3) status or merge with a larger, existing organization? In which country would the organization be based? Should they move toward a paid membership model or remain open to everyone? 

To shape AI4LAM’s future, Potter asked four MSI students to do what they do best: conduct qualitative research and find solutions that center people’s needs. 

From past to present

Julia Cave Arbanas, Austin Book, Madison Haack and Cameron Robertson are specializing in library science and digital archives at UMSI. This fall, as students in SI 633: Assessment in Cultural Institutions with lecturer Anthea Josias, they learned research methods for evaluating cultural heritage collections and services — and they went beyond theory, working directly with their client to make specific, actionable recommendations. 

“With the way that AI has grown as part of the cultural zeitgeist and exploded in use in everyone's lives, I think it's really had the folks at AI4LAM thinking about the future of their organization and what their role could be in driving the conversation,” Arbanas says. “This project represented an important opportunity to gather information before taking their next steps.”  

AI4LAM has experienced consistent growth — with as many as 400 attendees joining some new member meetings — but it’s challenging to understand the perspectives and needs of a dispersed and expanding membership. This is where the UMSI team came in. 

To gauge member attitudes, they designed a Qualtrics survey and analyzed responses from 44 AI4LAM members in 10 countries, who represented a range of institutions and professional roles. They also conducted qualitative interviews with three long-standing members.

“There seems to be a divide between people who like the more casual framework and those who would like to see a more structured organization,” Haack says. “By doing the interviews, by doing the survey, I hope we can give them a foundation or at least a snapshot of what their membership is feeling.” 

As a final but crucial step, they conducted a landscape review of similar organizations, analyzing their membership, activities, goals and leadership to help inform AI4LAM’s future structure. 

The goal, Haack says, was to give the client a path forward — “Not just to say, ‘Hey, here’s all this information, now go figure it out,’ but to say, ‘Here are some options for you if you would like to move in the direction of becoming a more official organization.’” 

To embrace new possibilities, you first have to understand the present landscape. 

“I think it’s going to be a really practical starting point for us to figure out how to make this organization real, and more sustainable and responsive to what members want,” Potter says. “The students at UMSI are just high quality. You can put them in a situation, and they're going to know what to do. They'll be professional. They’ll ask good questions.” 

This might be the most important tool for shaping a future where technologies like AI have the greatest benefit, and it’s a tool that information professionals wield well: asking good questions. 

LEARN MORE

Client-based courses provide UMSI students with high-impact learning experiences. Students gain confidence communicating with a client, while developing deliverables they can include in their portfolios. Learn more. 

Organizations from all industries and sectors are invited to propose information-based projects for students to work on through client-based courses, capstone projects and community-based programs. Host a student project.