What does your library need to harness the power of AI? This show is the first of a two-part series on AI with Christoper Jimenez, Web Services Librarian, and Melissa Del Castillo, Virtual Learning & Outreach Librarian, both with Information & Research Services at Florida International University Libraries. On this show, Christopher discusses Harnessing the Power of AI through Literacy and Frameworks.
Transcript
Library Leadership Podcast is brought to you by Innovative. Innovative, a part of Clarivate, is a globally recognized library industry partner with nearly five decades of experience developing library management solutions, discovery tools, marketing and communication services, and digital resource management products. Innovative believes every person in every community deserves a personalized library experience. Learn more at www.iii.com
Adriane Herrick Juarez:
This is Adriane Herrick Juarez. You’re listening to Library Leadership Podcast, where we talk about libraries and leadership, and speak with guests who share their ideas, innovations, and strategic insights in the profession.
What does your library need to harness the power of AI? This show is the first of a two-part series on AI with Christopher Jimenez, Web Services Librarian, and Melissa Del Castillo, Virtual Learning and Outreach Librarian, both with Information & Research Services at Florida International University Libraries. On this show, Christopher discusses Harnessing the Power of AI through Literacy and Frameworks. Enjoy the show!
Christopher, welcome to the show.
Christopher Jimenez:
Adriane, thank you for inviting me to your podcast and engaging with some of my recent musings on AI literacy. It’s going to be fun!
Adriane Herrick Juarez:
Question #1: In this episode, we hear from you, Christopher, about harnessing AI through literacy and frameworks. Then in the next episode, we hear from your colleague Melissa Del Castillo about harnessing the power of AI through collaboration, community, and communication. It’s a great two-part series, and I hope our listeners will tune in for both episodes to get a broad perspective on how they can harness the power of AI in their libraries. Let’s jump right in. Christopher, will you please define AI literacy? 01:31
Christopher Jimenez:
While library professionals can work together to highlight various aspects of literacy that could improve their learners’ understanding and implementation of generative AI, I like to use a definition of AI literacy that includes three discrete concepts. One—we want to teach the competencies. Two—understand how AI tools work that leads to critical evaluation and reflection on the tool, which ultimately—three, helps our students to use the tool to meet their needs effectively, efficiently, and ethically.
During our AI tools workshops at the Florida International University Libraries, we try to incorporate these three concepts into our lesson plans in hopes of improving AI literacy for our learners. These concepts foster transferable skills to any generative AI tool, whether it’s ChatGPT, Google, Gemini, Microsoft Copilot, Claude, DALL-E, Midjourney, and so on and so forth. We all know that trope about handing a kid a fish versus teaching them how to fish. Well, we want to focus more on the broad skills that help students navigate the world of AI and less on learning how to use this tool, or that tool; click this button; click on that button or whatever. So that’s our goal there.
Adriane Herrick Juarez:
Question #2: You use the Association of College and Research Libraries (ACRL) frameworks to create instruction strategies for AI. What are these? 03:21
Christopher Jimenez:
In general we use the Association of College and Research Libraries framework for information literacy for higher education. I know that’s a mouthful. Just call it the ACRL frames—to inform all of our instruction.
When I teach an ENC 1101, let’s say to students to help them to write a comparative analysis paper, I’ll use the frameworks to make sure I’m getting my point across in a way that not only helps them complete that one project that’s coming up next month, but also communicates the broader values of library research and authoritative information effectively.
So these frameworks have been around since 2016 and they are in no order: authority is constructed and contextual; information creation as a process; information has value; research as inquiry; scholarship as conversation, and searching as strategic exploration.
Of course, we try not to belabor the point or cover every single frame in full detail. Each class is unique and the instructor, that is the librarian, should only present the appropriate frame. I emphasize certain frames when I’m teaching ENC 1101, other frames when I teach a graduate level history workshop, and a completely separate set of frames when I teach citation management workshops. So it’s only natural for us to look to the frameworks for guidance when developing a workshop on AI tools.
Adriane Herrick Juarez:
Question #3: What are the layers of these ACRL frameworks? 04:54
Christopher Jimenez:
I also want to reemphasize what I said earlier, about how different library professionals can work together to highlight various aspects of literacy. The framework, again, is not an ordered list where you begin at one end, you work your way down the line until the end and poof, you’re done, right? I may interpret and categorize the frames differently than someone across the country. Heck, I might perceive the frames differently from my colleague who’s two doors down the hall. We’re free to stitch these layers together to equip learners with the information literacy skills that they need to navigate the world around them. So with this caveat out of the way, here’s how I layer the frames when teaching AI literacy. Again, we’re relying heavily on our definition that seeks to teach competencies, to understand AI that leads to critical evaluation and reflection on the tool, which ultimately helps our students to use that tool to meet their needs effectively, efficiently, and ethically.
So top layer, we’re presenting information to help our students understand how AI works. When we say how AI works, it’s not on a programmatic level. We’re not coders here, but we want to learn this at a level that demystifies the tool and sets realistic expectations for it. This relates to the frame information creation as a process, which highlights how the process of information creation—even when it’s AI, affects the final product.
A middle layer tackles the critical thinking elements of our definition when we ask how it fits into the research process. We’ve got Authority is Constructed and Contextual, which highlights concern about sources of data that AI taps into to generate their response. Also, the information has a value frame as the range of value added questions, such as the value of fast information or the value of accurate information, and how we balance the two within context.
Then a bottom layer or a wider layer, interprets how to use part of our definition in one of two ways. There’s what can be done, and then there’s how it should be done. The Research as Inquiry encourages the researcher to ask questions and let their questions lead to more questions. We just keep snowballing questions until we finally get something good to write about, or to learn about, or to tell other people about. We never just settle for that first answer that comes our way. Scholarship as Conversation invites the researcher to listen to multiple voices and to find their own voice within that scholarly discussion.
Then finally. Searching as Strategic Exploration encourages the user to be flexible and learn new tools, terms, tactics, and so on. So all three of these layers encourage the use of AI for exploration and inspiration. Yet at the same time, we caution users against pitfalls of over reliance and the uncritical use of generative AI.
Adriane Herrick Juarez:
Question #4: The library world is really working to understand AI, so knowing those layers is helpful. What are some other considerations that factor into your lesson plans. 07:52
Christopher Jimenez:
It’s such a great question. Thank you so much. Information literacy is actually really complex because we’ve got these literacies that stack upon each other. I usually make a point of emphasizing this fact that AI literacy requires one to also be, at a certain level, data literate, media literate, and digitally literate. It’s something that learners intuitively understand when you explain it to them, but they don’t consciously consider this. Rather than just tell them what I just told you, Adriane, I typically like to use this example—if I told you that according to the Small Arms Survey, there are 120 civilian-owned firearms per 100 people. You could do the math. You can estimate that this equates to about five guns per household—and the average family of four, right? Seems like a bunch of firearms per family, but this number is useless without context. This is one aspect of data literacy. We need to realize the population is diverse. We could parse this in terms of hunters, recreational users, and so on. We could use the rural versus urban divide. We can look at political ideology. All these different kinds of frames or ways—lenses through which we can look at this and interpret this. That’s a major tenet of data literacy, understanding that raw numbers need context to convey meaning. It also acknowledges that bad actors can manipulate context to weave a narrative, and not necessarily the bad actors part, but this is what the media likes to do. They give us a lens through which we can be led to interpret data, and this data has now been transformed into information.
So we hear this information from an MSNBC segment on gun control. It comes across very differently than if the same segment information was cited on Fox News during a segment on inner city crime, let’s say. So knowing that media conglomerates are often beholden to their shareholders, which can then color their presentation of news, again, it’s part of media literacy.
Yet again, we keep going. The medium through which we receive this information compounds these issues. So ask yourself, Am I reading this information in a long form piece of investigative journalism, or am I consuming a quick media hit that’s been formatted to fit into a social media clip? One of these formats gives you time to develop a complex perspective, and the other might just be rage bait, right? So in this digital world, it’s important to be aware of the complexity of our digital media.
And now, right now, is the point in which we introduce ChatGPT. What do you think the generative pre-trained model has been trained on? Guess what? The algorithm parses the data, media and digitized content that our society has created. So we think about these competencies. We think about these literacies as layers of a cake maybe. We take a look at that cake. We can tell it’s a delicious cake. It’s a useful cake. It’s going to be great. Right? We may not be bakers ourselves, but we should know the flavors of each layer, and the same is true for generative AI. We should use it critically and apply our information literacy skills to this informational tool.
Adriane Herrick Juarez:
Question #5: You also say AI is a progressive skill set that includes demystifying, demythologizing, demonstrating, and democratizing. Let’s go through those starting with demystifying. What is this about? 11:14
Christopher Jimenez:
We start by teaching that AI literacy is a demystified skill set. So by that we mean to say that AI is not magic. We type these words into ChatGPT, and they’re not just like incantations that bring about some sort of sorcery. There is actually a scientific model that explains what generative AI is doing. We can use that information, that knowledge we know about it to harness the tool to achieve certain goals
Essentially you learn that the generative AI is autocomplete on steroids. It’s not actually thinking through problems and then producing solutions. It is, though, rather good with words, and what words kind of are typically used within the context when you’re talking about a given topic. Then it generates logical sequences of related words on that topic. It also pretends to be an authority figure, so it’s going to be very straightforward like, This is the answer. So know that that’s the perspective or the tone that it’s going to take.
We teach our students to carefully examine how they describe their topic so the algorithm knows precisely how to respond. You’ve got to be explicit. You’ve got to tell it exactly how you want that information to come back to you. It may sound like alchemy, but generative AI does need those explicit commands and will only do what you tell it to do. And in that way, generative Ai is actually quite approachable.
Adriane Herrick Juarez:
Question #6: What is Demythologizing? That’s a tongue twister. 12:53
Christopher Jimenez:
Yes. And that’s one of those words that we trip up with all the time. Demythologizing. How do we even write that? Did I spell that right? When we talk about Demythologizing AI we’re talking about the upcoming AI revolution, right? We’ve all heard about it. Depending on who you listen to, AI is going to either usher in a utopia or a dystopia. Our robot servants are going to do all those daily mundane tasks so we can have the time for poetry, pottery, art and vacations, right? And alternatively, the robots are going to achieve the singularity, become self-aware, of course, decide to overthrow humanity and subject us to perform the menial tasks while their superior algorithms create art and go on amazing vacations and make pottery. So the major part of this skill set is just to be aware of these extremes and perspectives. Recognize that there are potential benefits and dangers inherent in each perspective.
Then we cut through that AI hype so we can enter into AI realities. Our understanding of how AI and LLMs operate—that’s large language models. I don’t think I said that explicitly, and as well as how we currently can use them is closely linked to the balance between hype and reality.
Adriane Herrick Juarez:
Question #7: Good to know. Will you explain about demonstrating. 14:15
Christopher Jimenez:
This is the star of the show. This is the part of the workshop that everyone envisions when they sign up, or when they ask a librarian to present AI tools for research. They want to know, How do I use the thing to get the thing done? Learners sign up for workshops in hopes that they can get a tour of the product, and a few tips on making the most of their time with the tool. That’s a critical aspect—that’s really important. They need to see it. They need that demonstration. They need to have that so they can go out and do it themselves.
However, as I hope I just demonstrated, literacy is much more complicated than just performing an action on a website and getting a reaction from the system. Once the components of AI are demystified and demythologized, we’re ready to demonstrate how to get what you want from AI.
So we demonstrate it. What can AI do? We discuss tools, writing tools, brainstorming tools, image generation tools, tools for various stages of the research process. Prompt engineering usually takes center stage as we demonstrate how quick off-the-cuff questions can be shaped using the CLEAR framework, which stands for concise, logical, explicit, adaptive, and reflective. This includes setting personas, audience, tone, and expectations appropriately in your prompt to get the best possible outcome. And yes, we usually do this live, in-person, and yes, we’ve demonstrated both the successful voila moments where everything works out great. The crowd’s amazed. They’re clapping. It’s amazing. My colleague Melissa, she’s had so many accolades for doing such great work.
Then also we’ve demonstrated those crash and burn sessions. When some recent update just changed the way the tool behaves and all we can say is, Wow, that was a major fail. Let’s try it again. So even in failure we can demonstrate the frameworks because searching for information is strategic exploration. Something fails, we don’t sit and pout. We ask ourselves why it failed, Oh, the tool changed? How can we adapt our search strategy to fit the new paradigm? Or I could have described what I needed differently, So let me try it again using these words instead of those words, or maybe be more explicit, or maybe I’ll pare this back—ask for something broader. So it’s important to demonstrate AI’s role in the research process. Actually, it is much more important to demonstrate the role than it is to demonstrate AI as a one-stop shop, right? Because AI is really not a one-stop shop for scholarship. It’s about demonstrating the process, not necessarily showing off the tool as a magic bullet.
Adriane Herrick Juarez:
Question #8: I think it’s good to know that the process is not always smooth. We must develop these skill sets to make this as seamless as possible. Finally democratizing—what does that mean in this process? 17:06
Christopher Jimenez:
This is actually a big question that extends beyond our classrooms. I like to think of the demonstrable skill set and the democratized skill set as two sides of the same coin. Where we shift the discussion from what can be done with AI on the one side of the coin, to what should be done with AI—or how we should go about doing it on the other side. So remember, literacy is always more about society than it is about academia.
We have ethical concerns we need to address when teaching AI literacy. These concerns can be put on display right alongside our demonstration. So as we’re going through—this isn’t like a new section that we say, Okay, we’ve done the demonstration part. Let’s move on to the next part. No, we try to weave these discussions into our demonstrations.
We talk about confirmation bias. I just did a crude example about the Small Arms Survey. We can look for opposing perspectives. We know in a classroom we want to encourage viewpoint diversity and critical thinking. So these questions that we pose to AI can be either consciously biased or unconsciously biased. We can infuse our questions with our preconceived notions. So we’ve got to pause to think about the difference between the two to help our students to recognize perspectives in what they’re reading and what they’re asking. Sources and misinformation. We talk about grounded AI versus ungrounded AI. Grounded AI will provide links to key sources in their response.
So remember, generative AI is programmed to bring back a confident sounding response to whatever is asked. It has immense resources via the LLM to generate that reply. A responsible user is going to double check the source to make sure that what ChatGPT or other generative AI tools are saying and make sure that it’s truly what the source is trying to convey. Sources and citation. Another hallmark of the responsible student is to give credit where credit is due. If the student wants to make use of a genuine point, that student will need to track down the source of that point and give credit to that person who generated that point, not necessarily the tool.
We also talk about business motivations behind ChatGPT. Things like privacy—how much of your information is given to these companies who take the information through ChatGPT, or whatever, and feed it back into the LLM to refine and improve their product? What are some of the steps we can take to protect ourselves from the more egregious incursions into our rights to privacy?
Another issue that I’ve recently been thinking about is how generative AI is also not as efficient as traditional search algorithms sometimes. When you go to Google, you run a search on Google. Gemini is going to pop up whether you want it to or not. Embedding AI into our Google and Bing searches—it’s not as useful as it appears sometimes, because it’s weaving a narrative when sometimes all you need is a link. That’s what you’re looking for.
But you know what? Our queries are helping those companies as they continue to improve their products. Then how are these companies profiting? Are they selling your information to third parties, or are we just using the free model to train up the tool until it’s good enough to sell these tools—are now twenty bucks a month? Who can afford these tools now? Who’s going to have access to AI at that rate? Who’s going to be left behind? How does this contribute to the digital divide? The haves and have-nots—who’s going to have the next necessary skill set to leverage AI and be competitive in the current, because it’s happening now—and in the near future workplace? Questions like, Did you know there’s an environmental cost to running server farms that support generative AI?
These are the questions we know our students really aren’t asking themselves when they interact with ChatGPT. They’re like, How is this going to work for me? How is it going to benefit me right now in the here and now? So just giving them a moment to reflect on some of these ethical questions raises awareness and helps place these issues in perspective. And that’s what we mean when we talk about AI literacy as a democratizing skill set. We’re trying to give our learners the tools to evaluate and use this tech as responsible citizens.
Adriane Herrick Juarez:
Question #9: And these are important things to know as information professionals—librarians, so we can share that knowledge with our communities. I really want to take your class. Is there anything else you’d like to share? 21:35
Christopher Jimenez:
The first thing would be that we record our classes, so you could probably just go on to our website, library.fiu.edu and search for the AI Tools Workshop. You could probably just watch it on YouTube or join us for a future session. I’m sure we could work that out. I’d also like to share that my work is not done in a vacuum. The frameworks for information literacy are almost a decade old, and they’ve been adapted to many layers of literacies and this discussion over the years. So this recent iteration for AI literacy stands on the shoulders of timeless legacies of information literacy, which is critical for our current society. Absolutely.
Locally, our AI Tools for Library Research Workshops are the result of discussions, planning, teaching, learning, revising, reteaching, iterating, and even more discussions. We’ve linked arms with—and linked brains with our other colleagues in the library, colleagues in the division of IT and other places at the university. So just know that there are experts all around you, and we all need to have the professional humility to ask for help and to learn from others.
Adriane Herrick Juarez:
Question #10: That’s so sensible. We all can benefit from asking questions, learning from others, and not being ashamed to say, Can somebody help me with this? I need some more information. That’s really wise. Christopher, do you have any favorite leadership books or resources and why? 22:55
Christopher Jimenez:
I’m embarrassed to say that I’m not really big on reading leadership books per se. I think I did StrengthsFinder once, it was pretty cool. I’m not really into reading those self-help books that are classified under HM 141 or whatever. But what I do like to do is read about a leader. Whether it’s someone who sets a positive or a negative example and learn some of the details of their lives, like how they approach problems. That could include presidents—I’ve read Meacham’s American Lion, the Andrew Jackson book, or Chernow’s biography of Grant.
Recently I’ve been fascinated by the life of Dietrich Bonhoeffer, Lutheran theologian who, despite his avowed pacifism, participated in a plot to try and assassinate the Führer, Adolf Hitler. There are a lot of books on his life, despite his untimely death at a young age, he was a very prolific writer himself—Bonhoeffer was. His primary sources are actually pretty accessible, including his letters from prison. I would recommend reading that if you want to learn from history.
Adriane Herrick Juarez:
Question #11: Thank you. I think it’s so important to learn from the lives of those who came before us. In closing, what do libraries mean to you personally? 24:18
Christopher Jimenez:
I’ve always valued free inquiry and the curious exploration of ideas. And there’s no place quite like the libraries for this kind of activity. Right? And yes, I mean, primarily access to the authoritative resources you can get on the shelf. But that’s not the only thing that makes a library one of the best places to freely explore ideas. I remember working in the library, in my library as a student, listening to the librarians engage with learners—helping them to think through projects and open new lines of inquiry. Not to mention when the librarians talked among themselves and figured out all the world’s problems, right? I mean, I’m just kidding, but really, if you just listen to them and take their advice, we’d be in a much better place. This is the kind of work that inspired and continues to inspire me to adopt a growth mindset and continually seek to expand my base of knowledge and build a web of interconnected information that I hope to use to inspire the next generation to engage critically with the information that completely envelops them. So that’s what libraries mean to me.
Adriane Herrick Juarez:
Question #12: Nice. You’ve done that on this show today, talking about harnessing the power of AI through literacy and frameworks. Our listeners will significantly benefit from this conversation. Thank you for creating a web of interconnected information we can use to better engage with and utilize AI. 25:38
Christopher Jimenez:
Thank you so much, Adrian, for inviting me and giving me some time on this platform. I really appreciate you and all the things that you’ve done.
Adriane Herrick Juarez:
Thank you Christopher.
Adriane Herrick Juarez:
You’ve been listening to Library Leadership podcast. This is Adriane Herrick Juarez. For more episodes, tune in to Library Leadership Podcast.com, where you can now subscribe to get episodes delivered right to your email inbox. Our producer is Nathan Sinclair Vineyard. Thanks for listening. We’ll see you next time.
We would like to thank the Park City Library for their dedicated support of this show. The opinions expressed on this show are those of the speaker and do not necessarily reflect the views of Library Leadership podcast or our sponsors.