Discussing artificial intelligence tools with information seekers: insights from practice

Background: This poster grows out of failure: my attempts to persuade researchers to use active machine learning tools for screening in systematic and scoping reviews. The novelty of AI in evidence synthesis methods is a real and perceived barrier, but other applications of AI already exist and are adopted with enthusiasm by information seekers. In this poster, I explore AI tools relevant to medical library users, including innovative tools (summarization engines and citation sentiment analysis), tools so common that we hardly think of them as “AI” (relevance ranking, article recommendation engines, and automated indexing recommendations), and tools relevant to evidence synthesis.

Description: With the ultimate goal of demystifying AI, I explore tools including Scholarcy, Scite, ASReview, ContReview, NLM’s Medical Text Indexer, Embase’s automatic indexing algorithm, and AI features embedded in databases including PubMed, Scopus, and Web of Science. For each AI tool, I describe the purpose, target users, ease of use, limitations, and potential value, based on my experiences using these tools and discussing them with students and researchers. These examples will demonstrate that AI is not the future, but the present; and that library users can come to trust librarians for user-centered advice about how to use AI tools effectively.

Conclusion: The buzzwords deep learning, machine learning, and AI surround us, and libraries are creating novel applications such as recommendation engines and chatbots. But while a few librarians are building AI tools, many librarians are giving library users advice about using AI tools -- some familiar and some innovative. In the context of reference, liaison, and research support, librarians have opportunities and responsibilities to help information seekers choose and use innovative AI tools, to recognize AI tools that already feel familiar, and to evaluate the familiar tools as well as the innovative ones.