Editors’ note: Today’s post is by Jane Jiang, the Director of Libraries at a community college in New Jersey.
For the past two years, I have watched conversations about artificial intelligence (AI) move from curiosity to urgency across higher education. The questions come from every corner of the campus community, but they often share a common concern. Faculty members ask how AI should be addressed in teaching and assessment. Students ask whether AI-generated information can be trusted. Administrators consider how institutions should prepare for a rapidly changing technological environment. Publishers are thinking about how scholarly content will be discovered and valued when researchers increasingly interact with AI systems rather than traditional platforms.
At first glance, these may appear to be separate conversations. They all point to the same fundamental issue: how do we preserve trust, judgment, and scholarly integrity in an AI-mediated information environment?
As a library director, I believe this is where academic libraries have a critical role to play. AI is changing how students and researchers discover, process, and use information. However, rather than making libraries less relevant, AI has made the human expertise of librarians more important than ever.
The future of academic libraries will not be defined by whether we can replicate what AI does. It will be defined by how effectively we help our communities navigate a world where information is increasingly generated, summarized, and delivered by machines.
The Research Journey Has Changed
For decades, academic libraries have helped students move from a research question to credible evidence through collections, discovery systems, research guides, and instruction. That starting point is changing. Today, many students begin their research journey with AI tools.
This is especially visible in community colleges, where many students often balance work, family responsibilities, and/or the challenges of returning to school. For many students, the hardest part of research is simply knowing where to begin. AI can lower that initial barrier by helping them brainstorm ideas, clarify concepts, and organize their thoughts before engaging more deeply with scholarly resources.
That accessibility matters. The question is not whether students will use AI — the question is how we help them use it responsibly.
One interaction shared by one of our librarians illustrates this shift: A student came to the reference desk after using ChatGPT to understand her topic and develop an outline for a paper. The response she received was polished and convincing. Instead of simply accepting it, she asked a much more fundamental question:
“How do I know if this is actually right?”
I appreciated that question because it reflected exactly the kind of academic responsibility that we hope students develop. She understood that AI could assist her work, but AI could not take responsibility for the accuracy, quality, or integrity of that work.
The conversation quickly moved away from AI itself and toward evaluating evidence, verifying sources, and understanding the research process.
That is where librarians continue to add value.
From Information Access to Information Judgment
Information literacy has always been about more than finding information. Students need to understand where information comes from, evaluate its credibility, and use it responsibly. AI does not replace these skills—it makes them even more important.
Organizations such as the Association of College and Research Libraries (ACRL) have long emphasized the importance of evaluating authority, context, and evidence. In the AI era, students must apply those same principles not only to information itself, but also to the systems that generate and deliver it. UNESCO’s recent AI competency framework similarly highlights the importance of critical evaluation, ethical awareness, and understanding AI limitations.
In other words, AI literacy is not separate from information literacy. It is an extension of it.
At our reference desk, librarians increasingly see students arriving with AI-generated summaries, research ideas, and citations. The challenge is often not whether the information looks polished, it is whether students understand where it came from, whether they can verify it, and whether it truly supports their work.
The role of the library is increasingly helping students move from simply receiving information to understanding and evaluating it.
Citation Literacy in the Age of AI
One of the areas where AI has intensified existing challenges is citation literacy.
This issue did not begin with generative AI.
Long before ChatGPT, libraries introduced students to citation management tools such as NoodleTools, Zotero, and other platforms that help organize research and create properly formatted references. These tools save time, reduce formatting errors, and help students manage increasingly complex research projects.
However, they also revealed an enduring lesson: automation can sometimes hide understanding.
A student can produce a perfectly formatted MLA citation without understanding why attribution matters, how scholarly conversations are built, or why researchers acknowledge the work of others.
When students transfer to institutions where different tools or expectations apply, some discover that they learned the mechanics of citation without fully developing citation literacy.
Generative AI increases this concern.
Students can now generate citations, summaries, literature reviews, and even research questions almost instantly. But a polished output does not necessarily represent scholarly understanding.
The purpose of citation has never been formatting alone. Citation is about transparency, intellectual honesty, and participation in a larger conversation of knowledge. The concern is not simply whether AI produces incorrect information. The deeper concern is whether students understand how knowledge is created, evaluated, and credited.
Academic Integrity: Moving Beyond Detection
One of the biggest concerns we hear from faculty is academic integrity.
A practical question comes up repeatedly:
How do we know when students are using AI appropriately?
Before generative AI, plagiarism detection already presented challenges. Faculty had tools that could identify some forms of copied text, but those tools were never perfect. AI-generated writing introduces an even more complicated situation.
Tools such as GPTZero, ZeroGPT, and other AI detection products have entered the market, but none can guarantee accurate identification of AI-generated text. Researchers have also raised concerns about false positives, and the reliability of AI detection tools, especially across different writing contexts and student population. For example, a comparative study by Weber-Wulff and colleagues found significant limitations in the accuracy of several widely used AI detection tools.
As a result, many educators are beginning to rethink the idea that technology alone can solve academic integrity challenges. The conversation is shifting from:
“How do we catch students using AI?”
to:
“How do we design learning experiences where students demonstrate their thinking?”
That means more emphasis on research processes, drafts, reflections, discussions, and transparency about AI use.
AI does not eliminate the need for academic integrity. It requires us to define it more thoughtfully.
The Human Role in an AI Research Ecosystem
There has been understandable concern about whether AI will eventually replace parts of library work. From my perspective, that question misunderstands what librarians do.
Students still come to our libraries. They still use live chat, email, consultations, and in-person assistance—sometimes even the day before an assignment is due.
The questions have changed.
Students increasingly ask:
“ChatGPT gave me this answer. Is it accurate?”
“These citations look real. How do I know they actually exist?”
“How do I find the original source?”
These are not questions that can be answered through retrieval alone. They require conversation, context, and judgment.
AI can summarize an article, but it cannot recognize that a student misunderstood an assignment after a brief conversation. AI can recommend sources, but it cannot mentor a student who lacks confidence in academic research. AI can generate text, but it cannot explain why one source is more authoritative than another.
Those remain human responsibilities.
The Human Layer
Libraries and publishers have always shared a common mission, even though we approach it from different directions.
Publishers curate, review, preserve, and distribute scholarship. Libraries help people discover, evaluate, interpret, and use that scholarship responsibly.
AI does not change this relationship. It makes it more important.
The future of scholarly communication will not be determined only by the intelligence of our machines. It will depend on whether humans can continue to exercise judgment in an environment where information is easier than ever to generate.
The most important questions remain:
Where did this information come from?
Can I trust it?
Why should I believe it?
Those have always been library questions. They are increasingly becoming publisher questions as well. As AI systems become new gateways to scholarly information, publishers will also need to consider how provenance, attribution, and trust are communicated when users encounter content outside traditional platforms.
The opportunity presented by AI is not to remove the human layer of scholarship. It is to recognize why that layer has always mattered.
Discussion
3 Thoughts on "Guest Post — The Human Layer: Why AI Makes Academic Libraries More Essential, Not Less"
It is absolutely true; the present generation is using AI tools in every phase of their research journey, right from conceptualization of ideas to summing up of entire research reports! Even the reviewers and publishers are leaning on reviewing the submissions using AI tools rather than evaluating the manuscripts with their intellectual abilities.
Is this the right direction the scholarly community is heading towards?
Honesty, trust, and research ethics carry and weigh high moral values in research endeavors on which the entire research community moves in a forward direction.
A single fabricated paper can contaminate the scholarly literature in the long run, and it is practically impossible to filter the residues to repair and eliminate such literature, as publication cycles are continuous, interlinked, interwoven, and spread across multi-domains and multi-directions.
Yes, the librarians, being custodians and caretakers of scholarly literature for centuries, know the values of refereed literature. And they are the able and reliable TRUST PARTNERS of researchers and academicians to reduce the damage AI tools may cause.
Jane Jiang’s argument that AI makes librarians more essential is persuasive.
Jane notes that provenance, attribution, and trust are becoming publisher questions as well as library questions. I would argue they have also become infrastructure questions, and that neither profession can resolve them one consultation at a time. The student who asked, “How do I know if this is actually right?” had a librarian to turn to. Most students posing that question to an AI system receive no answer, because the system retains no verifiable link to the source passage, the edition consulted, or the terms under which the rights holder permitted its use. The human layer Ms. Jiang describes depends on a trustworthy record beneath it.
Consider what a properly designed system would make possible. A student receiving an AI-generated answer should be able, with the publisher’s permission, to open the actual source document at the cited passage, browse the surrounding text, and generate a citation in whatever format the assignment requires. From that same point, she should be able to see which nearby libraries hold the work and where it can be purchased. This flips the LLMs from being a foe to a book promotion tool. Beyond verification, the publisher might authorize further interaction with the book itself—questions answered from its full text, for instance—under strict controls governing access, attribution, and compensation. None of this is speculative technology. It becomes reality when publishers and libraries unite to require it. The knowledge economy—publishers, libraries, and the scholarly societies that connect them—has the collective weight to define the terms on which LLMs and agents may draw upon its work. Individually, no press or library can impose conditions on a platform of that scale; collectively, they hold nearly everything of durable value those systems depend on. What has been lacking is a common position, and until one exists, the terms of access will continue to be set by the parties with the least stake in provenance.
To her three closing questions—where did this come from, can I trust it, why should I believe it—I would add a fourth, addressed to the systems rather than the students: can this answer lead me back to its source? I should disclose that this is the problem my company, Citations LLC, addresses, and readers may weigh my emphasis accordingly. But the alignment Ms. Jiang identifies between libraries and publishers is more than a shared mission. It is leverage, and it should be used.
Thank you for a thoughtful piece. You write: “AI can lower that initial barrier by helping them brainstorm ideas, clarify concepts, and organize their thoughts before engaging more deeply with scholarly resources.” I think this “barrier” is at least as important, if not more so, than those research steps that come next. These are the experiences, the challenges, we want students to learn and to practice on their own without the aid of articial tools – formulating ideas, brainstorming, early writing. Doing this with a friend, or a librarian (!), creates curiosity and engagement. The next steps are already aided by existing tools (databases) and these newer ones. And I want students to grapple with how the products of interacting with those tools leads to new thinking.
