Abstract
This presentation, given in December 2025, is a snapshot in time of Rutgers University Libraries' developing processes for assessing generative AI features in library e-resources. Over the past few years, e-resources vendors have been adding generative AI-powered features to library e-resources. In some cases, they are paid add-on features, while in other cases, they are added "for free" to a library's existing e-resources subscriptions. Options for enabling or disabling these features vary depending on the vendor. In this presentation, I conceptualize AI features not as replacements for traditional searching, metadata, and evaluation techniques, but as potential additional "tools in the toolbox" that researchers may find useful in their research process. I discuss my current process for assessing new AI features that are automatically added to our databases; the Generative AI Features in Major Databases & Platforms Libguide I use to communicate information about AI features to library personnel and Rutgers students, staff, and faculty; and my experience assessing and choosing to deactivate the Elsevier ScienceDirect Reading Assistant. It also discusses Rutgers' developing process for assessing more complex AI features that require deliberate enablement, including the AI Evaluation Rubric we created and how we used it to evaluate the Ex Libris Primo Research Assistant.