AI and the Library Reference Interview

More and more the general public is turning to Artificial Intelligence (AI) tools to assist them in their daily work and life.  AI tools draw on a large textual reference base (called a Large Language Model) to synthesize data and present it in a useable form.  Alternately, AI can take visual images and apply common forms of illustration techniques to come up with unique art.  Search engines are increasingly using AI to provide topical summaries to users.  AI is being embedded in our daily workflows. 

Ultimately, will AI tools take the place of a librarian reference desk interaction?  Or is the human component of this interaction still necessary?  Let's consider that question.

Engaged Inquiry

When a librarian interacts with a patron, they provide assistance.  It is an action of Engaged Inquiry.  How does a librarian provide this service?  It can be as simple as point a patron to key topical resources or research databases.  It can be as complex as answering specific questions or providing access to tools and equipment that the patron otherwise would have the means to utilize.

Frequently LibGuides have been used for answering repeat questions in academic settings.  They can provide a basic framework of historical, professional or community context so that a library patron can orient themselves to make sense of the information they are encountering.  Subject Matter LibGuides or Course-based guides can help support students and faculty in an academic setting.  In a Commercial setting Knowledge Bases or Corporate Wikis or Help Desk Systems do the same.  

Synthesis and Information Overload

Combating information overload is not just an analog task, it is a modern-day digital task.  Someone who is not familiar with the wealth of information available in a library may perceive the library as providing either non digestible information or too much low quality information.  Interaction with a skilled reference librarian helps in these cases.  In some cases, AI tools may also assist with this task.

Synthesis research and reports are used in technology or engineering fields to serve as a basis for answering industry wide questions.  Often these start with reference interviews and literature reviews.  NCHRP Synthesis Reports at TRB contain many examples of this type of research and reference work by senior researchers.  Many would argue that AI cannot take the place of the contextualization that these reports provide.

Should libraries and librarians be expected to provide this service? How is this service valued?  How much should learning assistance be provided to students or to the public?  Those questions move beyond the practice of librarianship and move into philosophy.

AI and Introduced Error

In 2025, when AI is still known to spew out results that are unexpected, nonsensical or erroneous, the usefulness of a reference interview comes into full view.  It becomes apparent that the who, what, why, and how questions guiding reference interviews are foundational to that interaction.  A team at MIT has documented instances of inaccurate content, hallucinations and bias in AI models and some of the real world impacts that have been experienced and provided strategies for mitigating bias and error when using AI tools.  It points to the value of the human touch in providing reference services.

A recent ALA presentation on Programming Librarian, Dr. Brandy McNeil listed several AI tools that she has utilized for various purposes.  These AI apps have a focus that AI models like Gemini, Co-Pilot and Chat GPT seem to lack.  Each of these AI models were designed with a particular function in mind.  

• Claude.ai - a framework tool for guiding data analysis
• Perplexity.ai - a trained ai that provides source material
• MeetGeek.ai - an ai tool for recording and reporting meetings

Other AI tools, are designed to take visual images and apply common forms of illustration techniques to come up with unique pieces of art.  A few of the apps I have briefly tested include:

• Meta's AI built into Facebook
• MidJourney in Discord
• Canva App

I found that each of these had artifacts or telltale marks in portions of the image that would point out to the trained eye that the images were not made by a human.  They had introduced error.  That used to be the mark of something handmade.  No longer.  Tech Republic has an article comparing the top 7 art-based AI tools and their costs.  

Provenance

The Tech Republic article says little about the fact that many of these AI tools have been found to strip artists of their intellectual property rights through reuse of their imagery.  Cases are making their way through the courts over professional usage of some of these tools.  OpenArt AI has 100s of apps designed to automate repetitive tasks for graphic artists.  However, commercial artists are still ultimately responsible for how they create their art and what reference images they draw upon for concept art.

Extending Limited Resources

Historically, the objections to a librarian providing full services in a reference interaction were twofold: 

• a full-service reference interaction robs the student of valuable basic learning experiences.

• a limited capacity to equally devote sufficient time and resources to provide full-service reference interactions to all library patrons. (the time cost of references services)

Theoretically, AI tools operated by informed users may have the potential to expand a librarian's ability to provide quality reference service work.  This is only possible if the AI tools are reliable and do not create more work for the librarian or introduce errors to the research and learning process.

FOCUS
A new computer I recently purchased came equipped with a "focus mode".  Basically, it is a tool used for ADHD students that sets a task timer.  The goal is to remain on task through the full focus period.  The computer assists in this process by limiting intrusive notifications and other alerts during the focus period.  It also assists in reminding a person to take a break when the focus period is complete.  It provides assistance in executive function monitoring that we used to expect of disciplined minds.

In today's world, people are increasingly delegating tasks to computers.  Even executive function tasks.  We are using the assistive technology tools we have created.  We need to recall that Focus is a component of attention and that ability to focus needs to be practiced so we can grown in that ability.

Synthesis is a necessary skill that should be taught as part of any profession.  It is absolutely a valuable and necessary service.  In some libraries, like upper level academic libraries this can be a useful reference interaction.  Sometimes it is senior professors or trusted researchers are tasked with this activity and it is those individuals who may chose to rely heavily on library personnel to distill and cull out redundant or dated materials.  NCHRP Synthesis Reports at TRB contain many examples of this work.  AI cannot replace these functions without loss.  Concise Library Guides

Draft 1: 8/10/2025