Sharpening ideas of AI for libraries at IDEA AI Institute | IDEA Institute on Artificial Intelligence
by Borui Zhang
Hi, IDEA blog site visitor! I’m an “AI librarian” I’m a human being too – sorry if I disappointed you. I work as the NLP specialist at my school library where we provide AI consultations, training, and resources to students, faculty, and staff. (“AI librarian” is the nickname my colleague gave me.) This is a new position in the library, and I appreciate that the IDEA AI institute 2022 fellowship opportunity taught me so much to help me shape my role in our library setting.
I very much enjoyed my experience at the institute. One of my favorite parts was learning about different AI aspects together as a team, where we actively exchanged ideas on what we learned from the sessions through hands-on exercises. The sessions were designed thoughtfully, and we were all able to discuss AI ethics issues in depth, and many of the conversations will be continually happening within the fellow community that we have created.
The role-playing exercise for AI research designing was a great learning experience and was also entertaining. It let us experience how important the roles (administrator, librarian, data scientist, engineers, etc.) are in an AI project as we played those roles. Understanding other roles’ perspectives is a key element of building meaningful products. A team with different professions does not necessarily guarantee a great final product until the team members truly learn from one another and elevate the group knowledge, skills, and experiences into the same task. Sudden situations (job change, family issue) challenge a project team from another angle. These scenarios helped us to set up a mindset of how to work through different kinds of difficulties by setting up realistic goals.
The technique AI trainings were highly practical. The instructors put great effort into laying out the history of those techniques, the theories behind the models, and letting us practice building our models with different platforms. There are different tools available to practice training models at different levels of learning expectations – some are more suitable for people who have less programming background to just grasp an idea of how machine learning models work at a higher level; some are involving codes to show what a dataset looks like, how it was passed into the model training structures, and how models are evaluated. These experiences helped expand my current strategies of how to select the right platforms and modeling tools for my patrons.
The future of libraries was another main topic that we discussed in many sessions, especially during the lighting talks and project presentation. We discussed many novel ideas for the kinds of AI-integrated services that future libraries should maintain. Some lighting talks highlighted the uniqueness of libraries — working directly with communities of different backgrounds and having a deep understanding of who we serve. Therefore, libraries should be involved in the development phases of AI applications that are designed to service large communities. It was inspiring of exchanging our action plans for the next steps of all the ideas at the end of the week.
Many many thanks to the organizers, the instructors, the fellows, and whoever made my experience as great as possible. I learned so much from all of you! Please reach out to me if you are interested in AI for library communities.