Generative AI Tools in Library Databases
The St. Thomas Libraries are trialing several databases in February, including two with AI-driven search interfaces: Scopus AI and Web of Science Research Assistant. The Libraries are excited about the potential for these tools to help researchers save time and uncover hard-to-find content. As we look at the broad range of AI-based tools that advertise similar functionality, we are closely evaluating their capabilities and differences.
What makes these library tools different from freely available tools?
The tools the library is trialing combine natural language AI search with high-quality, subscriber-only content. Both Scopus and Web of Science index over 20,000 journals, offering rich metadata that enhances summaries, relevancy rankings, and connections to related research. Additionally, as tools connected to our subscriptions, full-text access is often just a click away.
Free tools generate results based on either general web searches (Perplexity, ChatGPT Web Search, etc.), the Semantic Scholar corpus of academic documents (Semantic Scholar, Elicit, Consensus), or a mix of web content and publisher metadata they maintain themselves (Scite). These tools often do quite well at surfacing and summarizing relevant scholarship, particularly in fields with a lot of open access content, but results can vary widely depending on the specific subfield being researched. Some disciplines are well-covered and return excellent results, while others have gaps where important publishers or journals are missing. Free tools may also include predatory journals and student scholarship in the results they return.
So…is the library saying I shouldn’t use the free ones?
Not at all! We’re actively testing them alongside subscription-based tools to understand their strengths and limitations. Right now, it’s a “both/and” situation rather than “either/or”—free tools can help surface insights missed by traditional searches, but the tools we are trialing fill in many gaps left by the free tools, particularly when doing deep, comprehensive research.
We want your feedback!
We would love to hear from anyone who is interested in using AI tools to help them with research and has time to try out either Scopus AI or the Web of Science Research Assistant. If you have tried them, please take five minutes to fill out our feedback form. Faculty input is crucial when we evaluate new tools like this.