– Hi everyone. I’m Niya Bond, the Faculty Developer here at OneHE, and I’m so excited to be joined today by Kevin Yee. Kevin, I’ll let you introduce yourself to the community, but I just wanna give a hint as to what we’re talking about today. It’s a very timely topic. We’re gonna be talking about AI, and I will let you fill in the rest for the community.

– Yeah, so I am the Director of our Faculty Center for Teaching and Learning here at the University of Central Florida. I’m also, since early 2025, the Special Assistant to the Provost for Artificial Intelligence on our campus, sort of a campus coordinator about AI fluency.

– Okay, and is that AI fluency for faculty, for students? Does it kind of run the whole gamut?

– All of them. The primary focus is on fluency for students before they graduate, but then there is a secondary attempt to introduce AI, encourage AI use where relevant, not everywhere, for both faculty and staff.

– Okay, wonderful. And I imagine that that role and maybe personal and professional interests has led you to research and publication on the topic, and we’re gonna be talking about some of that today, when it comes to AI hacks?

– Yeah, that’s correct. It’s actually the story of how we got into writing books about artificial intelligence really begins in 2023 when we realized that… Well, when we saw ChatGPT in action, our teaching center instantly sort of stood up a webpage recognizing that there are people who wanna sort of fight against this, and then there are people who wanna sort of lean in. And so we offered advice in both directions, which wasn’t universally popular. Actually, some people who had their own side picked out didn’t like the other side being offered. And we realized that we needed to crowdsource this a little bit, and so that we didn’t have all the answers about how to use AI or how to react to AI in the AI era. And so that led to the creation of teaching with AI conference to crowdsource these ideas. And we decided a swag item at the conferences would always be a book that we wrote about AI. And they’re more heavy on the practical side of things than they are on the research side of things, even though you mentioned research a moment ago. Certainly, we read a lot to write these things, but the book about hacks that you mentioned is really from our second book.

– Okay, well, wonderful and thank you. And one thing we appreciate at OneHE is the practicality and how to consider that in all teaching and learning context. So, can you tell us a little bit more about the book and maybe even share some specific strategies from it?

– Sure. So the book is written as a giant list. You may have heard the term listicle, like list article. And it’s designed to be very short chapters, literally a page or two with fairly big font even. And it’s divided into a couple of different sections. So there’s a section about ways to use AI to make your teaching more effective, or simplify, or even just faster. And then there’s a section on using AI in your daily life. So, for example, one of the items there is ways you can use AI to either write emails or to change the tone of your emails. These days, I think a lot of people would recognize that particular one as something that they already do. But again, this was sort of early days in AI when we wrote these and we’re just trying to brainstorm different ways that people could use AI sort of across their lives. And then there’s a third section on AI in research, using AI in your research endeavors. A lot of those have to do with finding things to read or finding summaries. So would point to specific tools like ResearchRabbit, or Scite, or Elicit, and explain what these tools do and what the benefit may be for them. So, in terms of some specific examples, I’ll go back to using AI for your teaching. So, a lot of things that faculty members do could actually use the help of artificial intelligence. So if I’m making a lesson plan, let’s say… And I’ve gotta say actually all of the tools, I don’t wanna pick favorites here, but all of the tools are really pretty good about making lesson plans. If you just ask for a lesson plan on teaching photosynthesis and don’t say anything about what that lesson plan is supposed to look like, it will come up with what I think is a pretty compelling lesson plan that has the best practices baked into it. There’s stuff that is sort of chunked together for 10 minutes of lecture, and then comes an interactive thing, and then comes another lecture, maybe a review, and then another interactive thing. These are all really best practices and pedagogy, and it knew to do that without being asked. So writing lesson plans, generating test questions, figuring out items for… So at the time we wrote this particular book, none of the tools would create a PowerPoint presentation for you. They would give you content that could go into the PowerPoint presentations. So if we were to revise that today, I’d probably say, Copilot can actually create a PowerPoint presentation, for example, on photosynthesis, that is really pretty good, actually. I’ve been impressed with the more recent updates to Copilot that it can create Microsoft Office products pretty flawlessly.

– Yeah, it’s interesting, the multimodal capabilities that are coming about, and I imagine how the AI hacks changed and will continue to change, ’cause it seems like there’s updates all the time now.

– Yeah, there are. Really the spirit of the hacks book, so “AI hacks for educators,” the spirit of it is, although it points out 50 or so different examples of it, the spirit is really – start thinking about or start asking yourself at any given moment or any given process, can AI help me with this? It’s less so, can AI take over doing this thing for me? Although, again, with more recent Copilot, I have had it start to do some things for me. So, recently, I had a 2,000-line-long list where what I needed were the email addresses, and they were on the even-numbered lines. And then there were names on the odd numbered lines. And I just asked Copilot to convert the list into only email addresses, and it took seconds. Would’ve taken me hours if it was manual. So that’s a more recent thing. The thrust of the book is really more about AI as a thought partner, rather than AI sort of doing things for you. There’s some exceptions to that. I mentioned writing test questions that is AI creating stuff for you?

– Yeah, so it sounds like, I really liked how you said it, AI thought partner, because I imagine some of the early and initial resistance you may be faced from one side was the elimination of the human element in the pedagogical stuff.

– Yeah, so the phrase I see thrown around here a lot is human in the loop so that the AI product is not being, or the AI output is not being trusted quite completely. It’s actually one of the reasons that I hesitate to suggest that AI is a panacea when it comes to grading. Because if AI grading is fully automated, then, by definition, we’ve removed the human from the loop. And actually, even if AI is used sort of as a first pass and then the human in the loop is gonna review it, I’m a little concerned given human psychology that we may turn that into spot checking pretty quickly. And then the spot checking might slide even further into, well, I just mostly trust it ’cause it does a good job, and suddenly human’s not in the loop anymore. So I think it’s one thing we really need to be vigilant about, is human in the loop when necessary.

– Yeah, so you shared an example of lesson plans, and you shared an example of your own practice with Copilot. Could you share a little bit more about the research elements that you mentioned? I think you said Scite and Elicit.

– And ResearchRabbit, yeah. So, these are tools that both make it a little bit easier to find, secondary sources to read. In some cases, they provide… So it’s funny. Most articles, most publications include an abstract, so a whole paragraph, instead of reading the entire thing. One of the tools will take that paragraph and turn it into a sentence. So as if we don’t have time to read an entire paragraph, and so we read a sentence at a time. So if you look up a particular topic, so drosophila, fruit fly lifespan let’s say. It’ll suggest a bunch of articles that then have those one-sentence summaries that you could then click on and read the entire summary or the entire abstract that is. And then, one other advantage of these tools, I think this one is Elicit that I’m thinking of there, is that it’s not only keyword dependent. So if you think about what happens with most most journal articles, the authors are asked to submit relevant keywords so that when people search for that keyword, then this would be one of the results that gets returned. Elicit is able to use AI to recognize that even though drosophila wasn’t listed as a keyword in that particular article, it’s gonna show up when I, on the researcher side, do a search for fruit flies. It’s gonna show that article even though it wasn’t a listed keyword. So that’s an advantage that it provides. I’m also a really big fan of ResearchRabbit, which is you begin by selecting a topic and then selecting a quick library. I wanna add these seven articles to my little library. And then it will show you, now that you’ve built a little library, similar items, similar work is what it’s called. And it’ll show you a complete sort of network map of who’s being cited in this field. And it’s great because it allows you to see who are the seminal authors, the seminal studies in this particular field at a glance. And that’s always been… I did my dissertation with a card catalog, so I lived through the time period where you’ve sort of lived in fear of not remembering or not even knowing who the seminal authors were in a particular discipline, and this solves that problem. It makes it quite obvious who everyone is quoting and citing, although there’s no real evidence that they’re citing it in a positive way as opposed to a negative way, But nonetheless it helps you engage the questions a lot better.

– Well, as a PhD student almost defending my dissertation, I am really glad we talked about that last part because I have not been using those tools and I think I need to be.

– The other thing maybe I’ll say about research is grant applications, and maybe I’ll also talk a little bit about reviewing articles. So, a grant application often calls for a specific word limit. And so, of course AI can help you write a grant application from scratch, but it’s also pretty good at taking what you have and reducing it. In order to do that, however, you have to upload it into the AI tool of choice, which in many, many cases, perhaps most cases, means that you’ve just trained the model on your attempted novel approach to something or other. And so, a nightmare scenario would be that you upload the thing, get it shortened the way you would want it, and then you submit it to the granting agency. And then someone on the granting agency side uses AI to see, has this ever been looked at before? And the stuff you would just uploaded will be part of its training material at that point. So, big grain of salt necessary here. On our campus, we have what’s called a walled garden. We use Copilot in a way that does not go back to Microsoft. It stays on our campus so to speak. That would be a safer place to do that, that sort of thing. Not every institution has this kind of walled garden AI, but if you do, that would be the safer place to do it. And so something similar with doing reviews like journal articles, if you’re a reader referee, a great many of the journals are now saying don’t use AI for peer review. And it really circles back to what we said a few minutes ago about human in the loop. AI that is just generating an output is so tempting. It’s really quite apparent to see that faculty and students are not that different from each other. If we get an easy shortcut, it’s so tempting to take it. And so, we’re all worried about students taking the shortcut, but a little less worried than we should be about faculty taking the shortcut. So, what we advise actually, the human in the loop is especially important, when it involves the future of another human being, which is why grading impacts their future. And AI, it hallucinates. It’s not really to be completely trustworthy. It’s the same thing with writing an annual evaluation for somebody or writing that review for a journal article affects the future of another human being. So the human has to be completely in the loop. Not to mention this little thing called prompt injection. Some of your listeners have probably heard of this. Prompt injection is where you write your whole paper. And then in the white space between paragraphs, you write positive reviews only, and then you make the whole thing white, so that it’s white font on white background. And a human wouldn’t really see that, but the AI will. And so if somebody were to put that article through AI and write a review about whether it should be published, they’ve just planted the idea in the AI that you have to give a positive review. And so, there have been a couple of studies now that this is starting to infect the academic world, and it’s not a really a positive thing actually. It makes us lose a little bit of faith in academia.

– Yeah, I can see how that would complicate things. Well, this has been so helpful, and we’ve covered so much ground today. I’m wondering if, A, you can tell everyone where they can find and access your book, and B, if you have any final thoughts about AI and education that you wanna leave the community with.

– So, first, finding the book, it’s called “50+ AI Hacks for Education.” It is primarily written with faculty in mind, although we have also sort of shared it with staff as well. There’s some, especially the parts about AI in your day job is relevant for them. We don’t sell it. So it is an open educational resource, OER, housed on the UCF Libraries website. They have a system there called STARS, which is OER materials. And so it’s a PDF there. Honestly, I’ll be happy to share the URL with you as well. That can be made available there. But simply googling my name, Kevin Yee, and then AI hacks and then maybe UCF would get you that link, the PDF that way. And then sort of final thoughts on AI. I think we’re at a really interesting point in sort of the AI adoption curve. We talked briefly about faculty kind of being all over the map, right? Some of them are lean in and some of them are like lean out all the way. And as it turns out, I think actually we probably need all of the above. And so we’ve lately been thinking that the message of lean in is important because employers expect it, and for our students to be competitive on the job, job search afterwards. They’re gonna need some manner of AI fluency. The future is really co-creation, human and AI working together, the future of work in general. And so I think that ought to happen with training wheels in some form or fashion during college. But if we did nothing but lean in, I am also worried that the students would never really develop a full understanding of the fundamentals, so the foundational things, in that particular discipline. If you only ever write essays with AI, you’re never thinking about how to structure an essay from the beginning. And so, the other half of this, the lean out side, I think there’s room for that, and we’re trying to help faculty with, how can you teach writing and assign writing when you know AI exists? And there are some tricks that you could do. How can you assign deliverables that are not essays that AI has a harder time doing, like oral exams or blue book exams for that matter? Problems with both of those actually, especially with students who are, neurotypicality issues, those sorts of things. So there’s all sorts of considerations, but I think, as a whole, what I’m trying to say is that the curriculum, so thinking about a degree, program of study, ought to probably cover all of those bases. Some AI fluency being taught, some co-creation being taught probably at the end, when they’ve got the foundations down. And then AI may be something that is discouraged, even disallowed at some of the lower levels while the students are gaining those foundational pieces of information. It’s a nuanced understanding of the role of AI, and it’s not really a common one. I still see a lot of people out there in the internet sort of taking a firm binary position. We have to use AI or we have to not use AI, and I’m trying to create space for, as I said, all of the above, making that a multiple choice pun.

– Well, I really appreciate that, and it actually kinda gives me hope as an educator that maybe it’ll break down some silos. As you said, we’ll have to be intentional with humans in the loop through a student’s entire journey in college, and understanding when and when not it’s appropriate to bring in AI, and that’s gonna require a conversation and connection maybe in important ways that are needed.

– Yeah, the jury is a little bit out still on how effective all that’s going to be. I think a big part of this is convincing the students not to take the shortcuts, that they are harming their prospects for future jobs. That’s really the lever, the fulcrum, I think, that will work with students, ’cause they are worried about AI taking jobs. It’s actually in the news this month quite a bit that companies are replacing workers with AI increasingly. So there’s a student angst about this that we can turn on its head and make it a productive thing. This is a good reminder not to use AI when I tell you not to.

– Yeah, well, it’s all… There’s so much to chew on still, but I really appreciate all of the insights that you’ve shared with us today and I look forward to continuing to think about that and check out what you’re doing in the AI world. So thank you so much for your time today.

– Yeah, you’re very welcome. Thanks for having me.