Ben Lee teaches on the library master’s course at the University of Washington, including AI literacy, as well as editing a scholarly journal, Digital Humanities Quarterly. He trained as an astrophysicist, but got involved in human-computer interaction by working with archives, and seeing how librarians work with patrons to navigate digital collections in The Holocaust Museum. Watching people interact with card indexes to find out about their family led him to think how computation could provide new ways of accessing, and understanding, the archive. This led to a pioneering discovery system for exploring newspapers, called Newspaper Navigator, produced with the Library of Congress, where he was Innovator in Residence 2019-20. This discovery system included “visually similar content”, the ability to search via images, such as baseball players. Subsequently, he spent time as part of his PhD at the Allen Institute for AI, working with the Semantic Scholar team, investigating recommender tools.
Ben is someone with considerable experience of digital scholarly processes, and working with the next generation of information professionals, so what are his recommendations for the future? In a recent interview with him for the ATG Podcast, he came up with several interesting proposals.
AI Refusal can be a form of agency
Ben strongly believes the need to bring AI into the MLIS curriculum, but points out that AI refusal – the rejection of AI tools – should be taken seriously. That “refusal” might mean the rejection of for-profit, opaque AI models, for example.
Visual and multimodal search reveal hidden insights
Moving beyond text-based search to image, map, and other visual queries can uncover cultural and historical details that keyword search misses.
Minimal-resource computing makes AI more accessible
Using existing hardware and open-source models shifts AI from expensive, centralized infrastructure toward more sustainable and decentralized experimentation.
Conferences and groups help us evolve new solutions
Groups like CNI and ARL provide venues for us to engage in conversations about the best use of AI, and recognizing that it probably won’t just be one conversation that will lead us on the path here, but an ever evolving one
Our IP is being taken and scraped, no matter what we do
In our domain we have come to accept certain forms of AI already as in the norm. Ultimately we need to make sure the authors have a seat at the table and have an option to opt in or opt out.
AI literacy is becoming essential
Building AI literacy into education and practice helps practitioners evaluate systems critically, communicate responsibly, and adapt to rapid change.
I ask my students to create their own AI value statements
It is beneficial for students to articulate values that are important to them, such as transparency and accountability , and to demonstrate how they could apply them in the real world.
Open science counters proprietary AI dominance
Open science principles support more equitable, transparent AI development, and help reduce dependence on closed systems and private control.
I for one would be very pleased to have such a well-informed tutor on a library masters.
Leave a Reply