Hyperlinked Environments: AI Technology and Its Uses in the Academic Library
Hyperlinked Environments: AI Technology and Its Uses in the Academic Library
https://aiuniverseexplorer.com/wp-content/uploads/2023/10/2818e12d-010e-46cf-869f-e4991e6a9003.jpg
The more I read about AI, the more confused I get. At the rate that AI is changing, I am not sure I will ever get caught up. When I start understanding a few aspects, they change on me. As an academic librarian, I feel completely lost on how to train our students to use AI tools. I understand how to train the students to be responsible in their use of AI, but at the same time, AI is being integrated into so many applications that it is hard to keep up. At present, you can still choose if you want to use AI features or not, but will this always be true? Will AI agents become more of the norm rather than AI tools?
I gained understanding from Jeff Su’s explanation of the difference between AI tools and AI Agents. I now understand that AI tools need human input to get an output of information. AI Agents use data to make their own decisions on what input they want to gather and from where they want to gather it in order to get the output they want. AI Agents do not have human involvement (May 2025. Youtube video for Module 6 on Hyperlinked Environments)
As humans, we are always looking to make things easier. I see good and bad in using AI to make things easier. As examples of change that looked good but may not be so good after all, I will use calculators and cellphones. I struggled with the aspect of letting kids use a calculator for their homework when schools started that trend, because the kids are not learning to do the math themselves. They are letting a machine do the work, and they are staying in the dark as to how to do the math problem on paper and figure it out. There is so much accomplishment in doing the problem yourself and seeing it come out right. You learn from your successes and failures of doing the work yourself. I know people who cannot add and subtract on their own because they did not learn those skills, but instead used a calculator. I feel that using AI can do the same thing; it makes things easier, but also makes humans dumber. Another example is a cellphone. Cellphones are a blessing and a curse. They make us more reachable, but they have made us expectant of instant gratification. If we text someone, we expect a quick answer. I believe cellphones have easily complicated our lives and put unnecessary pressure on us that we are expected to carry. If you do not believe me, try going cell-free for a day and see what happens. We end up having cellphone withdrawal, wonder if we are missing something or someone, and we just plain feel lost. Then try that for a whole week, intentionally, and leave the stress of being on call 24 hours a day, and see how freeing that can become.
I know most of this has nothing to do with AI use in an Academic Library, but it does in my brain. If we try to make things easier for research, who really wins? Do we learn anything, or do we just let a machine do the work? Then the question is, can we trust the machine? It has been proven time and time again that ChatGPT makes things up – hallucinations and is getting better and better at lying and not getting caught (Mollick, 2023). I found this out firsthand while having to use ChatGPT for an assignment. I fact-checked what the output was from ChatGPT, and it was hogwash. I also found that you can ask a question more than once and every time get different answers. I did not know that ChatGPT was not connected to the internet (Mollick, 223), though, until I read that fact in several articles on AI tools. I guess the fact that we ask ChatGPT, which is on the internet, questions, I assumed the answers came from the internet.
So, let’s look at the use of AI tools in the Academic Library. Some libraries use Chatbots to answer simple questions. Some are using Chatbots to answer their phones (Cox and Tzoc, 2023). To me, these are both frustrating. I reach out for help and prefer to speak with a LIVE person. Chatbots can only give answers that have been programmed into them. Not all questions fit into the boxes provided. Most of my questions have not been answered by a Chatbot. Some libraries are using AI features in their online library card catalogs and databases. These are programmed on the Large Language Models (LLM) of the Libraries’ parameters, so they “should” be accurate, right? I am not completely sold on that yet. The Digital Theological Library (DTL) is working on an AI tool, and it is trained only on the resources stored in the DTL. This one hits home because it is Family of Faith Christian University’s largest online resource. I am still skeptical.
Those were tools used by the Library for their patrons. What about the AI tools patrons use in the library for their academic use? There are so many: ChatGPT, Dall-E, and Gemini, just to name a few. Each has its strengths and weaknesses. I am still lost on how to properly use them, but Cox & Tzoc and Mullick had a great list of what each can be used for and their good and bad sides (2023). I really would like to take a class on how to properly use them, where each is introduced and there are practical steps walking through their uses, showing their strengths and weaknesses.
On the flip side, do not get me started on the issues with students using AI tools for their writing and sourcing. There is an ethical way of using AI to help, and there is an unethical way. I believe if you use AI tools, you need to cite those tools and give snapshots of what you used and how you used what you used. That to me seems responsible and ethical.
Resources
Mollick, E. 2023. How to use AI to do practical stuff: A new guide. One useful thing. https://www.oneusefulthing.org/p/how-to-use-ai-to-do-practical-stuff
Su, J. (2025, May). AI agents, clearly explained. [Youtube]. https://www.youtube.com/watch?v=FwOTs4UxQS4&t=56s
Tzoc, E. & Cox, C. (2023). ChatGPT: Implications for academic libraries. Association of College & Research Libraries. 84 (3). https://crln.acrl.org/index.php/crlnews/article/view/25821/33770
5 Comments
Elena
Hi Maria,
Oh, how I resonate with some many things you wrote! I find AI one of the most dangerous tools available today. I expect the “danger” will change as I learn more and more how to better use it. I appreciated your comments about ChatGPT not being connected to the internet: that’s a fascinating “tip” to use when people say they trust using it for any matter of services.
I like to remind myself that AI and ChatGPT were the invention of humans who had to study and investigate and tinker that “old fashioned way” to create something artificial. So, too, students today need to study, tinker and fail and try again to create the next great thing. AI and ChatGPT are only as smart as we make them… this isn’t Harold from 2001 a Space Odyssey taking us over.
Finally, with regard to the phone and connectivity: everything you said is incredibly true. The device is a wonderful tool, but it is there to serve my needs. I cannot serve it. So I keep the ringer off. It take the pressure off, because I often miss calls and text because I am focused on my life and the needs of the moment: not on who might be calling. And while that sounds magical and unrealistic for many (I know!) I find the greatest panacea for too much of “too much” is working in the garden, or a long walk in a forest or beach. Disconnect from tech, and reconnect with me.
I was so glad to read your blog. We are in this together!
Elena
Jacque Fuller
Maria,
AI is so confidently incorrect, so people will believe its output. There are times it is helpful as an assistant, like in MS Word or Grammarly. Did you forget to attach a file? Would you like me to set a reminder? There are also environmental concerns, which some argue wipe out any ethical use of AI. The important thing is, do not use personal information with any chat AI. It is a large language module (LLM), which uses inputs to train itself and may keep information indefinitely. There are variety of videos that will break down how to construct prompts and parameters to get desired output. if you play around with chat AI enough you start to recognize patterns, especially if you ask it to meet word counts.
On a lighter note, this is one of my favorite YT channels. They go over the some history and patterns of AI videos: https://youtu.be/M4TXO4kQwSQ?si=V3RaOIBI5K9m-6lb
Maria Haeflinger
Jacque,
Thank you so much for that video. I was able to spot a few of the false videos, but was fooled by a few as well. The video was so informational and broke it down as to how to use critical thinking. I think I will use this for my next assignment.
@mariahaeflinger
Lindsay
Hi Maria,
You frame the complexities of AI in a way that I resonate with. It is meant to make our lives easier, but we need to stop and think that there are aspects of our lives that are not meant to be easy. Higher education is meant to challenge us. University students are meant to make mistakes and revise their work to learn what to do differently next time. If students are using AI to assist in paper writing, then the mistakes they are making are not their own; henceforth, what are they learning? In my opinion, your scepticism with AI in academic instituations is warrented because it threatens the critical thinking skills we need from students, ourselves included. Part of the challenge is that we are in what I view as a technology transition where we have this technology that is on the rise, but no one really knows how to teach it. What we can do is learn all we can about AI, test it out, talk about it, and come up with ways to explain how it can be used and more importantly how it should not be used.
Best,
Lindsay
Michael Stephens
@mariahaeflinger @mariahaeflinger, your mention of the DTL that is trained on the resources it contains is very interesting. It’s similar to something I’ve set up with my GPT that pulls from all of my published articles and books so I can ask it to summarize all the times that I’ve written about technology and academic libraries or something similar. If done correctly, the DTL may become even more useful to users via AI because it is tapping into its own data set.
You make a good point about a class focused on AI, and I believe those are on the schedule going forward. It might be very interesting to take one of them.