[Eril-l] Evaluating AI tools in licensed databases

[Electronic Resources in Libraries discussion list]

Hi Katherine,

This is something I am literally working on this week! While I doubt we would decide to purchase/cancel solely on the basis of AI, it seems helpful to have some framework within which to assess what the tools do.

Here is my very draft-y list of considerations:
---

  *   Interface
     *   How obvious is the AI function within the interface?
        *   How is it labelled?
        *   Where and when does the AI function appear?


  *   Entry points to AI function(s)
     *   When does the user engage with the AI function(s)?
        *   Can it be 'added' to the research process midstream, or must the user initiate research using the AI function(s)?
     *   Are there options to further engage with the AI function(s) as resources are discovered?
     *   Can the user disengage the AI function(s) during the research process?


  *   Help & Explanations
     *   How transparent is the use of AI?
        *   Can a user get information about how AI is being used?
     *   How naturally do explanations fit into the research process flow?
        *   Does the explanation open in a new window, in an overlay box, or with some other visual/layout - and is the method disruptive to the flow of research?


  *   Search Assistance - Natural Language Processing (NLP)
     *   Can queries be made in languages other than English?
     *   Can the user easily view how their query has been converted from natural language into Boolean terms?
     *   Can the user modify the Boolean terms, or remove the NLP step?


  *   Corpus
     *   What corpus is used for the search assistance?
     *   What corpus is used for summaries?
     *   Are some materials excluded from the corpus?
        *   Is so, which, why, and how does that affect the results?


  *   Summaries
     *   Are there topic or search-level summaries?
     *   Are there article-level summaries?
     *   Are summaries dynamic or static
        *   i.e. will the user get the same summary on a repeat search?


  *   Privacy & Reporting
     *   How is user privacy maintained?
        *   If the user is logged in, so they have an option to purge their searches?
     *   What statistics are maintained and available for library review?



From: Eril-l <eril-l-bounces at lists.eril-l.org> On Behalf Of Electronic Resources in Libraries discussion list via Eril-l
Sent: Friday, November 21, 2025 9:33 AM
To: Electronic Resources in Libraries discussion list <eril-l at lists.eril-l.org>
Subject: [Eril-l] Evaluating AI tools in licensed databases

EXTERNAL EMAIL: This email originated from outside of the organization. Do not click any links or open any attachments unless you trust the sender and know the content is safe.
Hi all,

I'm curious how folks are handling AI integrations in databases such as EBSCO, JSTOR, etc. How do you evaluate these tools, and how do you decide whether or not to enable them (when that choice exists)? If your library has developed a framework or rubric you'd be willing to share, I'd love to see it!

Thanks!
Katherine

--

Katherine Fish (she/her)

Serials and E-Resources Librarian

DeWitt Wallace Library 117

651-696-6530 | kfish at macalester.edu<mailto:kfish at macalester.edu>



1600 Grand Avenue

Saint Paul, MN 55105 USA



[https://lh4.googleusercontent.com/LweVzePvz8iDwSX0iIoZx9MkIi2jRdWpwSyTbkH8cpbf2tByHLByXpZZ7MYKpTPhHr2nbz-Sqnj9OF8pmmndnLL76kP5yinGsCzSQWJK2KF1fhVEpuJYXFsH2IMtUHu8eUMP3D-fFI1KtmBmVLD42w]<https://www.macalester.edu/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.eril-l.org/pipermail/eril-l-eril-l.org/attachments/20251121/c3a575f4/attachment.htm>