It's Academic

Like many library database collections, mine is centered around EBSCO. We have 28 different EBSCO databases for students to search within EBSCOhost Web, so any changes there have a big impact, and EBSCO has begun adding AI features to its products. There is a new “AI-Assisted Search” option that uses AI to enhance student search results. The latest try by EBSCO to have a Google-like search.

While that is interesting in itself, the features I’m more inclined to discuss are the AI Insights and Ask AI Features. Both of these offer enticing benefits to researchers while also having serious possible downsides for student learning.

Clicking the AI Insights link at the bottom of an article causes a summary of the article to be generated. The summary consists of four or five bullet points that cover key parts of the article. One would hope that these results are accurate because by integrating AI within our library collections we run the risk of allowing in the same inaccurate AI generate information we are trying to warn the public about. Libraries could become synonymous with “AI slop” if our offerings aren’t held to a high standard. I also can’t help but worry that students will avoid reading entire articles and instead rely on summaries. Real learning takes more than a few sentences. It takes some time for a new idea to cast its spell on you.

The AI Ask feature is more exciting. It allows the user to ask AI questions about the article itself. Is the methodology sound? What was the conclusion? How big is the sample size? I find this more beneficial to learning in that it requires the user to think actively to ask questions about the article. The AI becomes like a tutor explaining the article to a student. Of course, this does allow the student to bypass learning how to read complex texts. And the concerns about accuracy remain. Convenience has its price.

These new tools offer the same tradeoffs as other AI tools, convenience at the cost of trustworthiness. The possible replacement of human potential with AI. EBSCO allows these features to be turned on or off. Despite my misgivings mentioned above, I’ve decided to leave them on for now. It’s hard to resist the possibilities they bring.

As information literacy librarians, we’re accustomed to technology having an impact our teaching.  But gen AI has felt like a whole other ballgame with its rapid change and widespread impact.  Over the past few years, I’ve witnessed reactions to gen AI at both ends of the spectrum from outright rejection to uncritical adoption.  Where do we as information literacy instructors land on this spectrum?  And, most importantly, what do we teach our students about gen AI and the research process? 

In the Buddhist tradition, ‘the Middle Way’ refers to a balanced approach to life that avoids extremes in thought and action.  I would argue that many librarians teaching information literacy have been following a ‘middle way’ approach in response to new technologies for years.  With guidance from the ACRL Framework for Information Literacy for Higher Education, we teach about the authority concerns with Wikipedia while also explaining its usefulness as a research starter.  We explore the dangers of algorithmic bias in internet searching while also teaching students how to leverage Google results in the lateral reading/fact-checking process.  As information scientists, we are objective in our approach to technology and in our assessment of its impact on information and the research process.  We are not Luddites.  And we also don’t buy the AI hype.  I recently attended an ACRL CHOICE webinar,  “Genuine Learning in the Age of AI” and was encouraged to hear from scholars who are defining what it looks like to teach critical thinking and information literacy skills while using gen AI tools.  The panelists outlined approaches that emphasize critique, reflection, and reasoning while engaged with gen AI.  They advocated directing students to focus on the overarching question “Are you going to know when the AI is wrong?”   This emphasis on maintaining critical objectivity and developing information literacy skills while using gen AI charts a course for library instruction that avoids extreme reactions to this new technology.  

The Framework for Information Literacy was created to address the impact of emerging technologies on information and the research process and, even though written in 2015, it provides useful insight and guidance for analyzing gen AI tools.  The recent revision of the Framework for Information Literacy makes this application of the frames to gen AI tools even more explicit.  With the Framework and our perspective as information scientists, librarians are well-positioned to support students in the development of their critical thinking about gen AI tools and to define a ‘middle way’ for information literacy instruction.

As the Electronic Resource and Interlibrary Loan Librarian, I find myself being ghosted a lot.  And when I say “ghosted” I mean ghost citations and other AI hallucinations.  When I receive an interlibrary loan request (ILL), I always check the citation first for accuracy.  Thankfully, 98% of the time, these are correct and I can proceed with the request.  The other 2%?  That’s where the ghosts come in.

I’m thankful that ghost citations are not a huge problem at my university, however I see the potential for a host of problems in the academic world.  Over 10,000 peer-reviewed research papers were retracted due to AI fraudulency in 2023 alone (Yates, 2025).  If a trained researcher is having a hard time distinguishing real or fake information, how is a college student, new to research, going to know what to look for?   This is where academic librarians come in to save the day.  I like to think we have cool capes.

The world of research integrity is under attack from many fronts; we all see it in our everyday jobs.  Students are turning to AI to make their lives easier.  Research faster.  But, as information professionals, we need to teach them that easier and faster isn’t always going to give them the result they’re looking for, or at least what their professors want.  Going to college offers students the opportunity to be open to critical and creative thinking.  AI tends to dampen that.  A study out of MIT shows that 83% of students who use AI to help write their papers don’t remember what they read or wrote, because they lacked engagement in the assignment (More Curricular, 2025).

This next generation of students needs to be critical and creative thinkers, or the robots we fear may rule the world someday win.  We are human.  And humans need to use our brains.  Tease out the next problem.  Understand how to fix global warming.  Machines are not going to sustain us.  Only our living brains will.  I imagine our world becoming like the fictionalized story in Wall-E (which is my favorite Disney movie).  That story has the potential to come true, and that should scare us.  Living in a garbage infested environment.  No trees, no plants.  Living on a “cruise ship” in outer space.  Those people can’t even walk because they don’t use their muscles.  Seem familiar?  We can use machines to help.  To aid us.  But they should always be secondary to human thoughts, emotions, and creativity.

I’ll leave you with this.  We are at a precipice.  We have the potential to impact generations to come.  Why leave that to a machine?  Teaching our current students that they have the power to become amazing researchers, doctors, lawyers, scientists.  To use the beautiful brains we’ve been given.  What a waste to let AI “do the work.”  Teach our students to do the work.  To understand, to challenge what has and is being done.  That’s where innovation lies.  Let’s feed motivation to our students and give them practical tools to use in the real world.  Put on those librarian capes, challenge your students and watch Wall-E.

References

More Curricular (2026). The memory gap: why 83% of students can’t remember what they wrote with AI. https://www.morecurricular.co.uk/post/the-memory-gap-why-83-of-students-can-t-remember-what-they-wrote-with-ai

Yates, K. (2025). Citation cartels, ghost writing and fake peer-review: fraud is causing a crisis in science–here’s what we need to do to stop it. Live Science. https://www.livescience.com/human-behavior/citation-cartels-ghost-writing-and-fake-peer-review-how-fraud-is-causing-a-crisis-in-science-and-what-we-can-do-about-it-opinion

The development and capabilities of AI continue to grow faster than many of us can keep up. One new feature that I’ve found particularly startling is “agentic AI”. What sets agentic AI apart from a simple chat AI tool is its ability to actually complete tasks on its own. For example, you might ask an agentic AI to go to an online retailer and place a copy an item you specify into your cart. Asking the AI to “buy the best rated air conditioner on Walmart.com” and the it can actually manipulate your browser to do this. While this example doesn’t impact higher education much, it’s also possible to ask agentic AI to complete every quiz or discussion post in an online class. The student doesn’t even have to interact with the class in any way. I’ve tested this using an agentic AI browser called Comet Browser. With simple prompts on a chat window that loads on the right hand side of a browser the user can watch as the AI manipulates websites. The chat window even explains the process the AI is going through as it works.

Here is a video taken by a college professor demonstrating Comet Browser at work.

Tools like this will have a huge, almost entirely negative, impact on higher education. Any graded library content posted in an online class will be compromised as tools like these become more common. Academia will have to ask itself if graded online assignments are just a wasted exercise that proves nothing about student learning. The only possible solutions being proctored in person work or online proctoring tools.