Campus Technology Insider Podcast June 2025
Listen: Leveraging AI for Student Engagement and Success at Georgia Southern University
Rhea Kelly 00:00
Hello and welcome to the Campus Technology Insider podcast. I'm Rhea Kelly, editor in chief of Campus Technology, and your host. And I'm here with Scott Taylor, associate VP for student experience at Georgia Southern University, to talk about AI and student engagement — and ultimately student success and well-being. Scott, welcome to the podcast!
Scott Taylor 00:27
Thank you, Rhea. It's a pleasure to be here.
Rhea Kelly 00:31
So I thought maybe we could just begin by you telling me a little bit about yourself and your role at Georgia Southern.
Scott Taylor 00:37
Yeah, sure. So as you mentioned, I'm the Associate Vice President for Student Experience, and in that role, I provide oversight of the entire student lifecycle at Georgia Southern University, where we're a regional comprehensive Carnegie R2 institution. We serve over 27,000 students, including undergraduates and graduates.
Rhea Kelly 00:59
What led you to sort of rethink student engagement at Georgia Southern? Sort of, can you take me back to like when you were facing some challenges and what you were looking to tackle?
Scott Taylor 01:11
Sure, sure. So, you know, in the beginning, so to speak, you know, in my journey trying to support students, you know, the first thing that I can think of was working with students who are on academic intervention. We, you know, we're working with a population of students, trying to help them regain good standing. And as a part of that process, these students were meant to complete these academic improvement plans to sort of help them, you know, utilize resources, stay on task, and develop smart goals, you know, for the semester, and our engagement level was only 27%. So of the students who were supposed to complete these, only 27% were doing that. And I knew that part of it was we needed to engage in these with these students in a way that was more intuitive and more familiar to them. So we decided to adopt some technology to allow them to basically complete their academic assessments on mobile devices and to automate the process for the creation of AIPs. And so by altering the way that we engaged with students and making it sort of more compatible with, with the tools that they were used to, we were able to increase, and this is, this is in a single year, we increased that AIP completion rate to 99%. which is, which is great, and we end up seeing the impact from that, because, you know, a year after that, we see a reduction of 48% in the number of students who were on academic intervention. So that was kind of the start where I really felt like we'd come across a strategy for engaging with students and leveraging technology to be able to increase student success. So that was kind of, that was kind of the genesis. Where AI really came into play for us was we recognized that we could help students to ultimately be successful all the way through their careers if, if we basically intervened early. So rather than having that reactive approach, we wanted to adopt a proactive approach to supporting students. And that started off with, with kind of a classic chatbot, right, that we used initially only to engage students who were new to the university, so our freshmen. Just as kind of a guide to answer some basic questions based off of a standard knowledgebase, right? Help them get around campus, know where resources are, things like that. And so we started at that point, and, and that, of course, has, being able to leverage AI, of course, that tool has, has evolved dramatically, and that's kind of, kind of where we're at now. We're in a space where we're using that chatbot not as a chatbot, but as an actual virtual assistant now that that allows students to not just get answers to questions, but to actually resolve issues, 24 hours a day, seven days a week. So yeah, so that's kind of, that's kind of where, where, where we started, and where we're at.
Rhea Kelly 04:15
So it's not surprising to me that students would be, you know, most comfortable interacting via mobile device. How would you characterize students' comfort level with AI? Like where, where are we with, with that?
Scott Taylor 04:29
Yeah, so, so students today are digital natives, so for them, it's actually their preferred modality of communication. So, you know, students can always come into someone's office. They can, you know, go and, and meet with somebody in enrollment services or advising. They can pick up the phone and call the university there. You know, those, those resources are available. But by and large, that's, that's not what happens. Our students are much more comfortable, you know, chatting, which, which they can, you know, so our virtual assistant is called GUS bot. And GUS is, is, you know, he's AI powered, and we, we work with, with Druid AI to develop that, but, but GUS can be texted, or GUS can be accessed through our website, like either one of those. And students definitely would rather, you know, text their questions or seek to resolve an issue via text than they would calling somebody or even showing up in the office. Now we still get phone calls, but a lot of times it's mostly from parents, as opposed to students. Parents, there's again, different generation, different, different preferred mode of communication.
Rhea Kelly 05:38
So as you started to roll out a chatbot and then evolve that into a virtual assistant, what kinds of things did you need to think through in sort of setting up those technologies?
Scott Taylor 05:51
Yeah, so we knew that the capabilities existed. We just hadn't really seen them in the higher ed space. So a lot of folks had chatbots that, again, were based off of that knowledgebase, right, built off of a series of FAQs. We wanted a technology that we could leverage, that would actually resolve issues for students. Some of the things that we had to keep in mind for that was, this is probably the biggest piece, was data integration, right? How does our virtual assistant know, you know, if I'm a student, how does it know who my academic advisor is, right? How does it know whether or not I have holds, and how to resolve those holds, right? So that's a lot more sophisticated than, than a standard chatbot, right? And it requires there to be some amount of data integration, and also requires there to be some workflows that are built out so that the virtual assistant, assistant can resolve those issues. So, so capability was, was, was a huge piece, you know, finding, finding a virtual assistant that could actually do what we wanted it to do, that had the data integration, right? So I'd say capabilities, data integration, and then, of course, security is a huge one. So, you know, as a public institution, right, where we have to be good stewards of information and information security, you know, we're bound by both state and federal laws and regulations about data privacy and how that information has to be handled and protected. So we had to make sure that the tool operated with, within, within those parameters. And so those were kind of our three, three main concerns.
Rhea Kelly 07:34
Did you have any data cleanup that you needed to do? Like when I've talked to other schools about similar things, there's that idea that has come up that is, like, you can have, you have data you think is clean, but is it AI clean?
Scott Taylor 07:49
Yeah, so in terms of the data cleanup, I'll tell you, we were, we were, you know, cautious as we launched the tool, right? So, you know, again, not a lot of institutions had, had deployed this sort of tool, you know, before us. So, so we wanted to be careful to make sure that the data was correct, to make sure that everything was accurate, just because we wanted to protect basically ourselves from, from students opting out, right? So, you know, students being digital natives, they're used to a particular feel for technology, and they have a certain standard for how it is that they expect technology to perform. So we wanted to make sure that the information that we were providing was accurate, sort of from a student satisfaction standpoint, right? So we did. We were very careful, initially, to go through, we provided a knowledgebase, and we fact checked everything. I mean, it was over, over 3,000 different data points that we went through to make sure that everything worked out fine. Now, our partners at Druid AI had told us that, you know, they could have the bot crawl the website and glean the same information. But, but us, we were, you know, out of an abundance of caution, we decided to go with that knowledgebase initially. Now I will tell you after the fact, you know, they did have the bot crawl the website and do a comparison. And essentially, we reached the same, the same level of accuracy. So, so we kind of unnecessarily made some work for ourselves on the front end. But part of that was, you know, that was, that was sort of, for us in the beginning, a necessary step of getting everybody to feel comfortable enough to buy in and to make this move to leverage an AI tool. I think now we're in a very different space to where, because the functionality has been demonstrated, because people are now comfortable using the tool, it's like we don't have to, we don't have to be as cautious when, when we're ready to build out a new workflow or, or test out a new, a new idea. I think people have, have greater faith in the tool now. But yeah, we did some, it turns out, unnecessary data cleanup on the, on the front end, just as a as a precaution. Now, we do have instances where, let's say, for example, you know, we had very strange here in the south, but, but this winter, we actually had snow, right? And so there were a couple days where, where campus had to be closed, and, you know, making sure that the knowledgebase has that information, right? So, so we do have the ability to add some sort of one-off bits of information, you know, so the virtual assistant is aware of things that happen in real time. And we've, we've had some conversations around, for example, you know, so GUS, our virtual assistant, is, is an authenticated user, you know, giving GUS its own e-mail, right? And then, sort of having approved e-mail addresses that GUS can receive e-mails from for updates, right? So that if, if an alert goes out, you know, regarding a snow day on campus, or, you know, as we're approaching the hurricane season, you know inclement weather, right? That GUS, who would read its e-mail, would be aware of those updates, because, because it would constantly be checking and would be attuned to things as they happen in real time.
Rhea Kelly 11:18
Oh, wow. So you have it so that any updates that it receives from an official e-mail would just be incorporated into the data that it draws from.
Scott Taylor 11:26
Yeah, so right now, what we do is, is we have a process that is, that is vetted for updating information to the knowledgebase, but the capacity exists for us to automate that by providing the virtual assistant its own e-mail and doing just what you're saying, like sort of having the pre-approved channels that it will actually listen to. Because, of course, you don't want anybody to be able to e-mail it and give it information that might not be accurate. You know, everybody gets A's today or something like that, right?
Rhea Kelly 11:57
Yeah. New snow day.
Scott Taylor 12:00
Yeah, right.
Rhea Kelly 12:01
I think it's so interesting that, that, that it was a concern that students might opt out, you know, if it doesn't meet their level of satisfaction. I'm just curious, like, what you've had to do to encourage, you know, to sort of publicize that it exists and encourage usage among students.
Scott Taylor 12:21
Yeah, so that hasn't been a problem at all, and part of the reason is, is it's built into the onboarding process for our students. So when students come to the university, they receive an introductory e-mail where it's just like, hey, I'm GUS. I'm your virtual assistant. You know, I'm here to provide you answers for all things Georgia Southern. You know, whether you're looking for resources or wanting to know what's going on on campus, you know, feel free to text me, and I'm happy to help, right? So there's this sort of introduction that GUS does, and students will, will basically engage with, with the virtual assistant and kind of test it out. They'll, you know, just sort of what, whatever they happen to be curious about. So we see a huge influx of volume when that message goes out to all of our incoming freshmen. And what happens is GUS's number just gets saved in their phone, and if there's something that's related to the university, that's just, they're just going to text GUS and ask. Because it's easier than them going on a browser, right, going into the website and looking it up or anything like that. They can literally just, just text GUS, and that's, that's just such an easy mode, and in such an intuitive and regularly used mode of communication for them, that we don't, we don't really have to push it. Once students have started utilizing that their freshman year, they carry it on to their sophomore to their junior year.
Rhea Kelly 13:45
Okay, so, what about things like mental health support? You know, is there a built-in sort of system for detecting when a student might need counseling or something like that?
Scott Taylor 13:58
Yeah. So, so there are different, you know, so there are different things that would, would trigger sort of intervention, right, with, with the GUS bot. So by and large, when it comes to, you know, giving that sort of, that sort of advice, and things like, like, like, we're not going to defer to the AI to provide like counseling services, right? But the virtual assistant will make sure to connect those students to the right folks based off of, based off of what their need is, and then also to, if there's, if there's an issue that is of concern, to basically ping to let, to let a human know that, that hey, there needs to be some, some, some intervention and some engagement with the student from, from a professional, yeah.
Rhea Kelly 14:47
You talked about workflows before, and I'm, I'm curious about who needs to be a part of, sort of plotting those out and deciding, you know, what is the tool going to do when it detects X, Y or Z? You know, and how does it trigger things in different departments, like, how, how did you sit down and make, figure out those workflows?
Scott Taylor 15:09
Yeah, so, so, a huge collaborative process, really, across the, the entire institution. And that, and that's part of the reason why this tool, from a student engagement standpoint, lives inside, you know, Student Experience. Because it really does, it transcends the, the entire student life cycle, right? It's like, it, wherever you're at in your student life cycle, this is, this is a tool that you're likely to leverage. So we collaborated across every single division and across every single service area, with a particular focus on those that were student-facing. And what we did is basically an opportunity analysis, where we sat down with teams and we said, hey, you know, what are, what are your most common questions that you get from students? What are your most common issues that you're needing to resolve for students? Like, where is it that we can sort of take these transactional pieces and resolve them for the greatest impact, both for students, but then also for staff that oftentimes are resolving kind of a consistent issue or answering a consistent question for students. So what we did is across all teams, across all divisions in the institution, is we came up with a prioritization list, and, and off of that prioritization list, one of the things that we also looked at was, okay, you know, in order to build out this workflow, you know, what systems does the virtual assistant need access to? You know, what is, what's the lift for developing this, right, for developing this workflow and getting it to a resolution? Because some things are easier to build resolutions for than others. And then, based off of that, we just looked at, well, another thing we did is we looked at how many students, right? What, what is the volume of students that would be impacted by this, this particular workflow or this resolution? You know, again, where, where is it that we can have the greatest impact? And that's kind of how we prioritized and started.
Rhea Kelly 17:11
How often do you need to revisit those and kind of look at what might need adding or changing?
Scott Taylor 17:16
Yeah, so, so with I, with our ITS Team, we get together every month, and we have a, what we call an ITS prioritization meeting, and we look at sort of where we're at in the development of current workflows that, that are being worked on, and whether there are issues or anything with, with, with workflows that already exist. And by issues, really, it's more like updates, right, if we've had a change in policy, or if we've had a change in procedure. And then really looking at what other opportunities have sort of come to the surface that we recognize. So, yeah, that's, that's a…. The thing is, is, and this is what people, I think, will probably recognize about AI, is it's constantly evolving, right? So even in the time that we started with this tool, which has been, you know, two years ago, right, the capabilities of AI have, have moved so far that now things that might not have been options two years ago are options for us now in terms of development. So it's not a we're going to build this, and we build it, and it's done. It's a, this is where we're at, and as capabilities evolve, as students needs change, right, like we're going to constantly be evolving the tools that we have to meet students' needs. And that's, I think that's just a natural process that really probably should exist in the industry anyway, right, where we're constantly going back and evaluating, what are the needs, what are the opportunities, and then, and then moving forward, that way.
Rhea Kelly 18:52
Has there been anything that surprised you, the tool is capable of doing to support students?
Scott Taylor 18:59
Actually yeah. So, so this is, this is a really, really interesting thing. So I mentioned that we have, we had the knowledgebase that we kind of front-loaded for the virtual assistant, right? And when people contribute to the knowledgebase, or we were providing some sort of update, that information is coming from my colleagues, a lot of them student-facing. Which, you know, higher ed, higher ed is full of empaths, right? And it's been interesting to watch the virtual assistant adopt this, this language, right, this, this sort of shared language in higher ed that is, it's very empathetic. You can tell that it's, that, it's strange to say a virtual assistant is approaching its support in a way that's very holistic, right, and also very just, just very like, like, you can almost feel the timbre of voice of colleagues, like in the virtual assistant. So an example I'll give you is we had a parent that had actually reached out via the virtual assistant that was talking about, you know, her husband had an accident with the lawn mower, and they had to take a little trip to the ER, and it meant that they were going to miss Operation Move-In and bringing their daughter to the dorms on campus. And GUS bot helped resolve the issue, provided information, told them where, that it, you know, is fine, you know, this is where your daughter needs to check in, they'll be able to give her keys and everything. But then it went a step further, and it was like, you know, I'm so sorry to hear about this, you know, I can imagine that, that this is, this is only adding to, to your stress, but, you know, we're here and, and you know, wishing, wishing your husband a speedy recovery, and we look forward to having your daughter on campus, right? And I can tell you that that's not in the knowledgebase, right, but, but there is this sort of the collective language of people who provide service to students in higher ed who had kind of these, this shared culture and sort of worldview about like, why it is that we work with and support students, and seeing that language start to get adopted inside a virtual assistant has, it's been, it's been really interesting and kind of, kind of super cool. But it was like, that engagement was, felt a lot more relational than it did transactional, right? And I think a lot of times we think of these tools as, okay, it's resolving this transactional piece for me, but in reality, it's evolved to a point to where there's actually a particular standard and quality of care that enters that relational sphere, and that's super interesting.
Rhea Kelly 21:46
It's almost like that's something that would be impossible to intentionally build in, but it just sort of organically grew out of the general touchy feelies.
Scott Taylor 21:56
Yeah, right, right. Like somehow, somehow the LLM like, gleaned from everybody's constitution or from everybody's contributions that like that, that's the tone, and that's the ethos with which like service is approached.
Rhea Kelly 22:11
So can you share any like next steps of what you're working on with updating the tool?
Scott Taylor 22:19
Yeah. So, I mean, so we, we try to think as far into the future as we possibly can when, when we're looking at these technologies. I mean, we really try to stay at the bleeding edge of what's available, so that, you know, we can, you know, leverage technology, data and our people in the way that, that will have the most impact for our students. You know, I'm really excited about the, the, sort of the new, I know it's a kind of a buzz, buzz phrase, but the agentic AI capabilities, right? Getting these different sort of disparate systems at universities to, to align and work together a little bit more seamlessly. Because there are great enrollment CRMs out there that exist, right? And a lot of times you hear, you know, everybody talks about things being AI-powered, right? And, and, you know, there are some tools out there where, like, there's some component of AI in them, but getting all of these tools, whether they're AI-powered or not, to work together in, in tandem and to work together seamlessly. I think that's really where, where the future is at. So, you know, the example I'll give is, is all right, so we use a predictive and prescriptive analytic tool in our university, where we can, with a high degree of probability, assess a student's risk for attrition, right, whether or not they'll be successful at the institution, and that's based off of some, some really good historic data. So taking that historic data and that predictive analytic tool and being able to use agentic AI to pull in the, the, not the student information system, but the learning management system, right, that brings in grades. So I'll give you an example, like D2L, like we use something called Folio, where, you know, the students' grades in real time during the term, as they're entered, are in that, that learning management system. So being able to pull in that real live data, combine it with the historic data from the predictive analytic model, set up automations for interventions, so that if, if the model identifies that early in the term the student's attendance, or their first, you know, assignment that they've turned in puts, increases their level of probability, automating engagement to that student, right? And whether that happens from staff, or whether that's from the virtual assistant reaching out and being like, hey, noticed you missed your classes on the second week. You know, you know, is there anything going on, do you need, can I help you with something, right? And the automation of those types of engagements and connecting students to the right people at exactly the time when they need to be connected to those people. You could go a step further, and I think, you know, coming from, you know, the early part of my career as an academic advisor and building student schedules, right, you know, with our Course Search, which is, you know, part of our ERP, and then student's program of study, which is part of the student information system, agentic AI — and in this capability, this already exists, just the pieces need to be put together, right — could go through, look at a student's program of study, right, identify what their optimal schedule is. But then agentic AI, using, going back to historic data, could help that student avoid toxic combinations of classes, right, two classes that in combination create a higher, you know, drop, withdrawal, or failed, fail rate, right? But then at the same time, it could go through and it could cross-reference the Course Search and help build an idyllic schedule for that student based off their needs. Let's say they're a student athlete, and they've got a schedule, right, cross-referencing high-risk courses versus program of study versus course availability against a student athlete schedule. Right? Being able to bring together and optimize that data in a way that we as human beings would not be able to do, right? And so I think there's a, there's a quality of care that we will see come out of us being able to optimize the data that we already have, the systems that we already have, they just don't quite, the pieces don't quite fit well together yet. And I think agentic AI is the, is the solution for that. So I'm excited to see how it is that that changes, changes the landscape. And I really do, I think that, I think the impact on that, it's going to be, is going to be huge. I think that the student experience that will exist, you know, for most institutions by, you know, the year 2035, right, is, we're going to look back on kind of what we do now and be like, gosh, how did, you know, how did anybody ever make it through, right? How did anybody ever graduate, right? But, but, yeah, so I'm super excited about, about some of those capabilities through agentic AI.
Rhea Kelly 27:30
Thank you for joining us. I'm Rhea Kelly, and this was the Campus Technology Insider podcast. You can find us on the major podcast platforms or visit us online at campustechnology.com/podcast. Let us know what you think of this episode and what you'd like to hear in the future. Until next time.