855: LIVE From Ellucian Live 2024 - with The Ellucian Team, An AI Discussion

Joe Sallustio: Welcome back everybody. It's your time to up on the EdUp Experience podcast where we make education your business. This is Dr. Joe Sallustio back on the third episode here on the first day. Where am I, you might ask? Well, I'll tell you. I'm at Ellucian Live in San Antonio, Texas. There's going to be an eclipse. It's going to be cool. The speakers are amazing. There are people everywhere right now. And I have just been ambushed. I think it's an Ellucian takeover on this podcast episode by five Ellucian employees, engineers, CISOs. I don't know what they do. There's a lot of brain power here. I'm by far the dumbest person around the table by far. I'm going to go around and I'm going to ask everybody to introduce themselves. I have an applause for each of these. So I'm going to go from left to right. Here we go. Who are you? Actually, my audience knows you. Go ahead.

Sonia Khan: I'm Sonia Khan. I am the VP of Mischief and Cookies at Ellucian, but my real title is VP of Innovation and Experience Design. Thank you for having us again and accepting this large party.

Joe Sallustio: So happy to have you here. Sonia has been my co-host now two or three times, I think. But this time you're guesting. You're a guest this time. Slash co-host. You can do whatever you want actually. All right. Number two. Here we go. Let's get the audience. Get back going here.

Lacey Gerard: My name is Lacey Gerard. I am Sonia's hype woman and I am a senior manager, experience design, and an adjunct faculty member.

Joe Sallustio: All right. Well, welcome to your first. How many podcasts have you done before?

Lacey Gerard: Approximately zero.

Joe Sallustio: So this is your first podcast. It will be unlike any that you do from this point on. All right. Number three.

Jen Welding: Hi, I'm Jen Welding. I am a deputy general counsel, vice president and chief compliance officer at Ellucian. So mind your P's and Q's.

Joe Sallustio: I will do that. I'll be very compliant around you, Jen. There we go. Number four.

Robert Levy: Hi. I'm Robert Levy, VP of engineering. I focus on our SaaS platform products and AI.

Joe Sallustio: All right. And you guys want to, as you talk into these microphones, this is OK. You've got to be about three inches away from the top so we get what you're going to say. And number five and final.

Josh Sosnin: Hello, hello. Josh Sosnin, Chief Information Security Nerd, or like you said, CISO is the real title and 0.0 podcasts for me.

Joe Sallustio: Actually, I don't have it right now, but I have a sound effect that says 0.0. I thought you might. Yeah, I didn't have it ready. So thanks for calling me out. You know, we're here, you guys, to talk about technology, compliance, ethical AI. There's a lot going on in the space right now for us that we don't understand how technology... I use AI. I use AI because I'm not skilled. But no. But what does that even mean? I use AI. How does it work? What are you doing with it? How do you make sure that it's compliant, that it's not going to give us wrong information? How do you embed it in products? I have so many questions. Lacey, I'm going to start with you just since you're directly across from me. Because you said user experience. And so that's what I immediately go to is how does this work for me? And should I even be worried that it's behind the scenes, right?

Lacey Gerard: There are so many great ways you can use AI in user experience design. One way we're seeing a lot of people use it is to kind of take the user research data that we have and generate insights from that. We also have a lot of human biases and using AI to kind of come back up behind you and make sure that you've gotten all the relevant insights without leaving anything behind because you got really excited about one thing that you heard. So that's one way that we've been seeing it in UX. Of course, there's a lot of crazy ways that people are using AI in design work. We're not really doing any of those right now. But yeah, in UX, it's great for kind of looking at our research and making sure we got everything.

Joe Sallustio: Tell them like it is. What's the group say about this? What's the group say about what Lacey said, user experience, user design with AI?

Robert Levy: I think the tools that are coming to the forefront around building user interfaces with AI are fantastic and very exciting, and we're seeing things like that in all kinds of spaces. When we're thinking about UX with AI, the most important thing to recognize is that AI is just based on probabilities. It's a mathematical model trained on whatever data was plugged into it that's predicting the future, right? The next word a chatbot should put in a sentence or the probability of a student being retained until next year. It's all based on probabilities. And that means that it's guaranteed to be wrong sometimes, which is something that is a challenge that we have to make sure we're accounting for in our user experiences.

Joe Sallustio: It's fuzzy math. But since you said it's math, I mean for me it's, I don't understand that. I don't understand math. How is it math and probabilities? I mean, are we talking about, that seems simple, right? Not that I would be, I'm not very good at math. I'm just gonna put that out there. But it seems simple the way you describe it to the user, it seems much more complicated than that. Like, how is this doing this? And, you know, whatever this is, there's so many "this's" that AI can do. How is it working? And is it all based on ones and zeros? You know, I mean, is it all come down to something like that?

Robert Levy: Everything comes down to ones and zeros. It's always about math. But it is fuzzy math. And how it actually works, how you get those ones and zeros is going to depend on the use case, right? The way that ChatGPT is coming up with its ones and zeros to predict the words is very different from the way that we would use ones and zeros to predict whether or not someone's going to be retained at a school the following year.

Sonia Khan: Remember the year that ChatGPT started to hit the world? Robert and I were at a Christmas party with our boss, Mike Wolf, Chief Product and Technology Officer at Ellucian, shameless plug. And I remember showing Mike what ChatGPT was doing. And Robert, you were sitting right across from me. I remember you looking me in the eyes and like at my wide-eyed excitement for it and saying, "Sonia, I promise you, it doesn't actually know anything. It's just predicting things. You can't trust it to actually do relevant work."

Robert Levy: Yeah, it doesn't know anything. It is making predictions based on what it's seen in the past, but doesn't actually understand. And that's why it's so important that we're very thoughtful in how we use the right kinds of AI in the right use cases and have the right user experience to establish trust and that we're following all of the regulations and best practices.

Joe Sallustio: So when I think back to many, many years ago, first year of college, many, many, many years ago. I had to learn how not to fail out. What about the flip side of retaining, right? Is there a way that ultimately we can help students figure out? So for me, I learned that I need to sit in the front row of every single class I'm ever going to go to from now on. And I didn't even have the distractions of cell phones and other things. Can you imagine something that's also going to help the students prosper?

Robert Levy: Yeah, 100%. And this is one of the things we're really excited about with our insights platform is we're bringing all kinds of data into one place. And once we have all that data in one place, we're able to have these very powerful predictive models to determine things like that. What advice should be given to a particular student? What types of engagement should be had, what are the great courses they should be taking, what's the right schedule for them. That can all be suggested and predicted by these models.

Joe Sallustio: I like your style dude.

Lacey Gerard: I think students that are using AI are doing really creative things with it. I'm teaching a course right now and there are four groups and they were supposed to generate user scenarios over the past week. One group generated one, two groups generated two, three scenarios each, and then one group generated five. Guess which one used AI to help them with their homework? It was that one. So they were able to really leverage the technology in order to blow me out of the water with their homework. It was amazing. I was very pleased for them.

Joe Sallustio: Was it good quality?

Lacey Gerard: We're at an early stage of the design work in this part of the process so it's okay that they did that. I did not, you know, I try not to look too closely at AI generated wireframes for example because they're not going to be like, they're great. But they needed to generate a lot of ideas and directions and this was like the perfect time to use it.

Josh Sosnin: So the killer app for AI is filling out all the HR forms and yearly reviews and self-evaluations is that it?

Jen Welding: That is definitely one of the things that can help with.

Joe Sallustio: Attorneys can use this like for briefs and stuff?

Josh Sosnin: No, Josh is kidding.

Jen Welding: You know, I think that's why it's important for there to be a balance between the innovation and then, you know, knowing what the guardrails are, following, you know, keeping up with the legal regulations and laws all around the globe to understand and make sure that the use cases are all consistent with that and within the guidelines. Really like the, who owns the data is kind of a big question that we grapple with with AI applications and so you know we've stood up a team at Ellucian. Robert's a big participant in that. We stood up a governance process because look, we know it's huge, it's an innovation we want to be part of it, we need to be part of it and we are, but we want to make sure we're doing it in a way that is compliant not, you know, using it to do HR reviews. Josh, I will say yes, don't listen to Josh on that one. Listen on other things that you talk about.

Joe Sallustio: So schools, right? And cause I work for one and we woke up at one point earlier in the year and went, a minute, we don't have a policy around AI. Our, some executives are just plugging in financial data and you know, sensitive information cause it helps synthesize some of the things, but then your name, your institutional names on it. You've just made all this information public now. If you look at the fine print in some of those programs, they own that information. It's like Facebook back in the day when you put up photos and now your photos are owned by somebody else. That's something that schools have to watch out for because it's really easy to go, let's upload our financial data that is very sensitive that we don't want anybody to see. And all of a sudden now it's public, right?

Jen Welding: Yeah, no, that's a great, you know, that was our, could be some corporate experiences too, or technology company experiences. We need to, you know, AI got out there and everyone was using it. And it was important to develop a policy to educate everybody on what that policy is, to have a governance structure, to have guidelines so that, you know, again, we want to balance the innovation, but we want to be doing it in a good way.

Joe Sallustio: Josh, is that a key for you? Like, I've got to clamp things down so information isn't going where it shouldn't be, but also have the balance of opening things up enough so that people want to use AI. It's a really fine line between how I use AI as somebody that doesn't understand really a lick of what all you are talking about, I just go, ooh, have you heard of this AI program? And I start plugging things into it. Versus student information, versus organizational information. How do you find that balance between we're using AI and we can't do some of this stuff?

Josh Sosnin: Yeah. We don't want to be the no fun police, but oftentimes we are, right? So hear it. Can't have no fun. The goal is to make sure that we can innovate and get some great features to customers and allow the customers and empower the customers to actually use them and not run into compliance issues and security issues. Now, I recorded you. I told you earlier that I recorded you twice walking by because somebody said something about AI and uploading sensitive data and I recorded you saying this. Execute Order 66.

Joe Sallustio: I don't know what that means, but you're going to have to fill us in.

Josh Sosnin: Yeah. So part of it is let's keep this secure so that bad things don't happen to your data or so that there's some crazy stuff you can do playing prompts and getting large language models to do things that they're not intended to do, so that's certainly part of it. But then the other side is we need to make sure that the way that we're doing it as a company is going to allow our customers, wherever they might be throughout the world, to actually use the features based on whatever regulatory and compliance requirements they have there. So it's just, we're really doing a lot around AI. I mean, we have partners like AWS and CrowdStrike, they're doing some amazing stuff with AI on the security front that are helping us. But then there's just a lot of work we're doing to make sure that our products are going to be able to be consumed.

Sonia Khan: Yeah. We actually published our data ethics standards back in 2020, right before COVID hit. And in that we said, look, we need you to trust us with your data. We are not going to use your data to benefit anyone other than you. We published that in 2020 and we're sticking by it and it's working.

Robert Levy: But the other thing I think it's important to recognize is that AI isn't, in terms of data protection, AI doesn't change the game, right? Should you be sending sensitive data to a third party? No. The fact that somehow people forgot it is like, You get it. It's like, but it's really fun. No, you still don't do that. That hasn't changed. But AI does introduce other types of risks around prompt injections, a whole new vector of security problems that can come from generative AI. If anybody, any of the nerds in the audience are familiar with SQL injection, prompt injection is basically the same thing, but fuzzier. And that can lead to all kinds of crazy things happening.

Joe Sallustio: Do you think schools are, Ellucian Partner Schools are looking to you all to help define what the landscape of AI is going to be for higher ed? You know, and as a person in an institution, I go, okay, who do I go to in higher ed to learn more about AI? Well, there's the companies like Ellucian that are leading the way. You have access to tons of data. You're EdTech, right? So you're not burdened by the structures of higher education because there are many and they are burdensome. So you can go out and do things that maybe a higher ed institution can't. What are schools asking you for at Ellucian?

Sonia Khan: Yeah, I think where Ellucian is unique is that we truly understand higher ed, whereas some of the other big players would be able to tell you how much power they have behind something. We have the experts who could tell you, let me show you ways that this could be ethically used to support retention. Or let me tell you other ways where we can check, potentially scan, dialogue, and tell you whether or not a student is showing signs of well-being issues. Like we can do things where we connect it to the real problems of higher ed, whereas I think other big players will just talk to you about power.

Robert Levy: We did a survey of all of our customers recently, that's published now, and it came back that everyone's very excited about AI, a bunch of them are using it, not using it in super complex, sophisticated ways yet, but they see the use cases they want to get there, and the things that are holding them back are questions about how the heck does this actually work, I'd like to understand the software I'm deploying. This is lunacy. Concerns about the potential costs, and concerns about unknowns of the risks that are there. So we spent a good amount of time talking about what those risks are because they're the same risks that we face when we're adopting AI internally and sharing our best practices. The reason you have such a big panel here, people coming from different disciplines is because that's how we tackle these risks.

Joe Sallustio: Well, and it all comes back eventually to students. I'm looking at you, Lacey, because I go, there are still some faculty out there that are not. And I say that, like still some faculty, like people should be surprised. It's not surprising. There are very slow adopters to the point where there are faculty out there going, if you use AI, you're just going to immediately fail my class. If you bring a calculator to the math class, you're going to fail math. I mean, calculators, my six-year-old has a calculator. You know, I mean, this is one of those times in our industry where there's a spectrum of adoption, right? There's an adoption. What do you call that? The adoption curve or whatever it is. But you can't stop the train here. I don't even think you can turn AI off if you wanted to at this point. It just doesn't turn off, right? There's faculty members who have been teaching the same class for, you know, decades. And it is hard to change. Change is just difficult.

Lacey Gerard: I try and use it. I mean, I'm kind of on my own as an adjunct. I have to make my own curriculum. I have to make all my own rubrics. I just have it get busy for me in ways that really save me time. Like, I do have it make my rubrics for me. I have it beef up my assignments when something isn't clear. I use it like kind of a shadow co-worker because it's just me working from home. So I use it. I like run things past it. I'm like, hey, how can I make this better?

Joe Sallustio: You're using it so you have more time to actually serve the student, right?

Lacey Gerard: Yeah. Like spending time on things like writing rubrics is not how I want to spend my weekends. I have three kids. Like I don't want to be writing rubrics all weekend. So I just use it to just try and like support me as a faculty member and it works.

Joe Sallustio: Well, what does this conference mean to you all here? What do you... When you guys go home at the end of this, I've asked this to the guests too, but for you all supporting schools. What does success look like at the end of this conference? How do you define whether this was successful for you, for Ellucian? I mean, it will be, but what are your takeaways? What are you hoping for?

Sonia Khan: I hope our customers see that we're innovating, that compared to a lot of our competitors, we have a crystal clear view as to what the ethical embracement of AI is, and that they see tangible things on our roadmap that are showcasing how we can do things like Lacey mentioned, which is use AI to supercharge their tools so that they can focus on student success.

Joe Sallustio: Anybody else?

Robert Levy: Yeah, as success, we have a vision of the future we're executing on. We have been executing on. We have a lot of awesome things that are now available and we want to see customers get excited about those things and start engaging and driving success with them.

Joe Sallustio: You're looking for compliance?

Jen Welding: Always. No, I mean it's a great opportunity for us to get in front of customers, to talk to them, to Robert and Sonia's point to show them what we've got going on, what the tools are that are available to them so that they can really focus on student success.

Joe Sallustio: Lacey? Success, what's it look like for you?

Lacey Gerard: I just want to see all the cool new stuff that we're gonna show.

Joe Sallustio: New stuff, that's it's all about. You gotta have some new... What I'm excited about. Epic. Josh, wait for it.

Josh Sosnin: I mean the people who get super excited about security aren't exactly here but there are some folks from a ton of different schools that I've worked with over the years where they do have compliance issues. And we've been able to help them, there's been a never-ending wave of new compliance issues on the security side that they've had to comply with and well, we can't make you compliant. I know our customers wish we could snap our finger, they become compliant. We do help them through features in our products. And we have, I would say, really gone to bat and delivered features in our products that have helped and certainly in some areas where they would have had to do it on their own, but we've built it into our product. So just kind of talking to some of them, seeing how that's been going and anything new that they've seen that we haven't heard about, it'll be great to...

Joe Sallustio: I love that you're sarcastic about security, I love it, right? Because schools are getting nailed now. I mean, you don't have to go too many weeks out to know, there was a breach in this higher ed institution, there's a breach in that higher ed institution. There's a real weakness in higher ed around security and... as the music turns up here at Ellucian Live, we're going to talk more and more. When I said security, the music came on. That's how much you should be paying attention to security in your institution. Guys, what an honor it was to have you in the podcast. I feel like you can, why don't you guys come back at the, on Wednesday or Tuesday night before I leave and just wrap it up for me. Like, did we hit all the things that you wanted to? What was the excitement like? You know, we can do a wrap up episode. All five of you see if we can do it again now that you're loose, you're looser. We'll get you looser on the second one. Ladies and gentlemen, see if I can get it. Sonia, Lacey, Jen, Robert and Josh, Ellucian smart people. That's what I'll call you, the Ellucian smart people. It's been an honor. Thank you all for being here.

All: Thank you so much.

Joe Sallustio: Ladies and gentlemen, you've just ed-uped.