This content is drawn from a report authored by the AU Library's Artificial Intelligence Experimentation Working Group. You can read the groups full report covering experiments with the use of AI in library work and recommendations to library leadership in the American University Research Archive.
This Working Group’s purpose was to identify and reduce inefficiencies in important library processes, gain deeper insights into ways to use AI to improve services, and optimize existing workflows. The careful and strategic integration of AI into AU Library’s administrative tasks, data analysis needs, metadata creation actions, and ticketing and support processes, generated new frameworks for understanding potential innovations in this space and laid the groundwork for future collaborative projects. Each subgroup’s work uncovered areas where AI might be useful in improving operations and that ongoing dialogue across departments will be essential to fully realize its potential on campus.
AI and AI integrations are still maturing, and future advances will open more paths for use. Since the experimentation concluded, some of the AI tools examined have added or improved their AI assistants. For example, Airtable has introduced an AI assistant, and Power Automate now has a ChatGPT connector.
In order to fully realize the potential for AI use at the University Library, the AI Experimentation Working Group strongly recommends continuing experimentation. Subgroups made great progress applying AI tools to library workflows and found ways to support internal work, but none of these experiments could be deemed “complete” at the end of the experimentation phase. The subgroups did not find any definite solutions, so the work with AI is not finished yet. Any future work with AI will need continuous experimentation. As such, it is important that a collaborative working group structure remains in place to guide the gradual adoption of these new technologies. One possibility is to create a standing experimentation group for work with AI. This would allow us to request funding for research and experimentation and could provide opportunities for library employees to publish or present their findings.
If an enterprise-level AI tool is adopted in the library, this Working Group also recommends setting clear limits for the adoption of AI. It is crucial to remember the amount of human labor that is needed to program, train, and prompt AI in order to achieve the results that meet library needs. Library operational needs are more niche than the AI can currently provide without significant human guidance and review. In this phase of experimentation, experimenters were unable to create any situation in which AI could completely take over a task and run on its own with minimal oversight. In addition, each new person that approaches the workflow would have to do almost as much work to get the same result from AI as the first experimenters did. Therefore, if the AU Library expands AI use internally, it will be necessary to carefully consider which tasks are selected to use with AI, instead of attempting to apply AI to all workflows.