The Conference Called EDUCAUSE: Reflections from EDUCAUSE 2025

Welcome to Educast three thousand. It's the most transformative time in the history of education. So join us as we break down the fourth wall and reflect on what's happening. The good, the bad, and even the chaotic. Here's your hosts, Melissa Lobel and Ryan Lufkin.

Alright.

Well, welcome to a very special episode of the Educast three thousand podcast live from Educause twenty twenty five in Nashville, Tennessee. We're actually recording this in our booth. Hopefully, sound quality is better than last year. You may get a little our podcast listeners were actually on mic in the booth as well as recording. So if you get some weird sound, we apologize. But yeah, I'm joined by my good friends and one of the smartest people I know, Zach Pendleton, Chief Architect here at Instructure.

I don't know how I'm gonna follow-up that intro, man. Thank you for having me. It's so good to be here. And, again, with you, I hope our sound quality is better, but I have high hopes that it won't be any worse than it was.

I don't think it could actually be worse than it was. So okay. So it's been most of the full day. We're in the afternoon or the first day of Edge of Cause, and there's some trends definitely teasing their way out.

One of those, obviously, is AgenTic AI. And there's news happening this week. There's some interesting things being talked about here. Give us a little bit of insight into what you're hearing.

Okay. Yeah. So, you know, at the beginning of the year, there were certain people who couched twenty twenty five as the year of the agent. Now I think those people are maybe a little, ahead of the curve.

I I don't think that we saw a lot of adoption earlier in the year, but we are seeing now this real push and excitement for Genic AI. So, you know, if we talk Genic, what we're really saying is a large language model that can do things for you Yeah. In the systems you already use. That's Canvas or something else.

But, you know, at this conference, I've had a lot of people talk to me about model context protocol Yep.

Which is an open standard that came out of Anthropic that makes building these types of agents easier. And I'll be honest with you, if you had had asked me about my bingo card for Educause, MCP was not on it. You know? But I I think that shows you that AgenTic is really top of mind for a lot of people there. And we're seeing, I I think some of that may be driven by the market deals that people like Perplexity, people like OpenAI are now releasing these agent powered web browsers.

Yeah. I think the thing with MTP that a lot of people are really interested in is it actually helps with security, right? It helps control what these tools have access to, especially in education when we talk about student data privacy and data security, that's a big challenge, right? So, and in an open architecture like Canvas, right, it provides the level of flexibility, but then an increased level of control and things like that, right?

Which I think is fascinating. And I want add to what you said earlier. I certainly think that twenty twenty four was the year of fear of AI and twenty twenty five is about putting it to work. So I think that agentic aspect really is about finding meaningful ways to apply this technology that really does save educators time, really does improve the experience for students, things like that.

Right?

Yeah. Absolutely. And I think, you know, I'm glad you mentioned security about MCP because I do think as we talk about all these new standards, there's this challenge to think that everything is brand new. Yeah.

And really, MCP, I think, so powerful because it builds on top of all of the investments that we've all been making in open APIs and in security over the last decade of end tech, which is great. You know? And I think, you know, as we look at broader adoption, man, I I hate to be the technologist who shows up. I I pride eyed and excited about the future, but I think there's a real potential when we think about agents now being attached to things like web browsers, inspiring a lot of really cool creative use cases for faculty, for students, for others who were programming may have been too hard.

Oh, you know, even MCP requires you to stitch two systems together. These browsers are like it's like Microsoft Excel for kind of general purpose learning and and browsing.

Yeah. Actually, and last week, was actually at ASU's Genetic AI and Student Experience Conference, which was all about the agent, right? And they showed off their Create AI, which is creates the ability for educators with no coding skills to be able to create agents to accomplish tasks, right? And I was really, really interesting because they're exposing educators to these tools to let them explore how to possibly use them the best, right?

Overcoming that fear, thinking of new use cases, those types of things. And I think we're gonna, you know, my mission really is to surface more of those because I think, you know, we've talked about this earlier today as well, but I think more educators need to have a little bit more what are these tools capable of? What does good look like? What are some of the possibilities?

It can be a little blue ocean, right? Like, and where do you start? And I think that's one of the cool things that's coming out of the conversation right now.

Yeah. Absolutely. I know you and I have had conversations about this, and I've had a lot of conversations with schools as well, but you referenced twenty twenty four being this year where people were kind of afraid of AI. Everyone was just kind of afraid to poke at it. Yep. And I think as people start to use it, they realize it's maybe not as scary as they first thought it was. And and they realize that it's a new tool, but it's not a goal.

Yeah. You know, I I don't think any classroom ought to take a goal to just use more AI. Yeah. But it does open up a lot of possibilities for instructors to change how they teach even if their goals are the same.

And so kind of this harmony, this rethinking some of our age old pedagogy Yep. That served us because it was the best thing at scale we had at the time, maybe kind of do for a refresh. And I don't know, I know you've had a lot of these conversations too. I would love to hear some of what you're hearing about AI pedagogy and how teachers are succeeding in applying this stuff.

Yeah, mean, there's some really interesting like scenario builders, right? I was talking to my friend Neil from Utah State University, and and he was talking about the tool they built, and he actually has given me access, but I have not had time to play with it yet. We'll dig into it together later on. But they actually worked with I mean, I love that he their professor Lee reached out to a dungeon master, D and D dungeon master, who is used to building those scenarios, that tree based kind of choice, choose your own adventure type of narrative, and helped them build the scenarios for a leadership training scenario.

And is personalized. The experience is personalized based on the choices the individual makes and AI creates a seamless path for them. And so I'm really interested in digging into that and really showing more of those. So again, I think seeing what they're capable of, they're not just chatbots, they're not just cheating tools.

There's so much functionality that you can get into it. And then as you see those, you can say, oh, you know what? I thought about doing something like this with that course, but I remember being able to do that at scale across an entire class because it would just be too time consuming. And then I can, in a lot of cases, actually save huge amount of times and scale those interactions.

I love that use case. I think that type of experiential learning and really brings a level of practicality into the classroom. Yeah. That may have been missing with other enlistment types.

I, you know, this actually reminds me of the report that we just released here at Educause. Right? So our readiness and and workforce report. I got the name of that wrong, Ryan.

Gotta tell me what the name of the thing is.

I believe it's the learning and readiness report. So, Yuriyo, are you there?

We talk in higher ed about how much students are using AI.

Yeah.

What this report showed us, when we talk to students who graduated, we talk to Gen Z. Gen Z is the generation most afraid of technology in the workplace. Yep. And so I think what we're seeing here is this world where you've you've got a generation that is so comfortable with technology in their personal lives, but is struggling to bridge that into academic and professional success. Yeah. And so tools like this that pull some of that experiential learning forward into the classroom seem like they make learning more relevant.

Yeah. Break down less barriers.

Yeah. Generating.

Yeah. Well, and and one of the other interesting stats that came out of the report was about only about fifty percent of the individuals surveyed were actually using AI on the job, which is lower than you'd expect, right? We have this assumption that like most businesses are deeply using this. And while, you know, we use it pretty heavily, we've done a lot of enablement and structure, You know, I would still say it's lower than we would anticipate AI usage by most employees.

And so seventy percent of students don't feel prepared for the workforce. Fifty percent, only fifty percent are using AI. So I think they're more prepared than they believe they are. And how do we give them confidence into moving into those roles and being able to like create their own agents, you know, leverage these tools?

And again, we're in the kind of throes of developing, I think, muscle around that, right? And the idea that we have the power to do that. Most businesses are looking for experts in AI to help advance themselves. And so these students have the ability to do that.

Now, I should say my daughter, who is a junior in college, she's of the kind of, I think, tranche of students who is scared to death of using AI in the classroom because she's been told it's cheating. And so she's not using it in the way that she should. Her little brother, that's six years younger, is using it in inappropriate ways, right? Like, he's out there, like, using it to the extreme because he's actually more savvy with AI than most of his educators in high school.

And so how do we actually create some consistency around that, a consistency of message? That's my biggest fear is that we've got this inconsistent kind of tranches throughout education, throughout age groups, and what's acceptable and how to use these tools. And I think the same applies to technology in a lot of ways.

Yeah, absolutely. I mean, I will say, you know, another wrinkle to that is the socioeconomic one. Yeah. That we end up in a world where students who come from backgrounds, where they did not go to schools that were maybe as as wealthy Yeah. Or as privileged Yep.

Tend to use less AI, and then they're even further behind the curve when they get to a spot where they're supposed to use it.

Yeah. But I think for me, you know what? For all of the fear and the uncertainty, when we look at what's happening in the workforce, we look at education. I think the silver lining here for me as a human is that this really reinforces to me how important it is for humans to have human conversation as and to decide how they can work together to make this stuff work for us.

Yeah. And I that's where I think there's a lot of opportunity for educators and for students. Right? Let's let's take people who know the tools.

Let's take people who know how to teach. Let's bring them together, let's figure this out together.

Exactly. Well, and then and, you know, employers into that, and say, okay, here's the skills we need. Let's really focus in on the skills we need new employees to have when it comes to AI or just in general, but let's suppose those loops and bring that feedback together. And I think the more we can do to create those conversation loops, the more, you know, we launched Canvas Career at InstructureCon, right? And the idea that we're, that's really a skills based learning tool for employers, but it carries on the experience within Canvas that our, you know, students love.

And, you know, how do we kind of continue to build that lifelong learning journey, support that throughout, I think is incredible because AI is going be a big piece of scaling that. And we see that divide with universities. Some have a lot of money to spend on this. Some are being very exploratory in how they're using AI.

Some are really, we talked to a couple earlier today that are not doing a lot, right, at this point. And it's because it feels overwhelming. Their faculty are maybe more hesitant to get engaged. But ultimately, you know, if we're not engaging with educators or not engaging with students, the educators aren't saving themselves time.

They're not leveraging the benefits of AI. And the students aren't leaving with the skills they need to be immediately effective for jobs that are used to that. Right?

Yeah. Yeah. Well, and I think there's this fear if you've not been using AI, you know, if you're one of these institutions that doesn't have an acceptable use policy or is is struggling to figure out what tools to use, you you feel like you're behind. And then I think that feeling and that fear leads you to do less.

Yeah. Because you feel like maybe you've missed the game. But I I mean, I think what I just want everybody to see is that, man, this space moves so quickly. Yeah.

Things that I was talking about eighteen months ago don't matter right now. Yeah. You know? And I sunk a lot of time into figuring those out.

So if anybody needs AI knowledge that's eighteen months old, I'm looking for a job.

I'm very happy with it. But I think, you know, there's no better time to just start using the tools. And research shows again and again and again that people who have the highest opinions of AI and the people who are least afraid of it are people who use it. Yeah. And I think that's because it just takes the magic out of it and it helps people focus on what actually matters.

Yeah. When you approach it from a collaboration standpoint and you realize, you know, Lane Freeman is with the North Carolina system who, Trinity College system, he's their head AI trainer, and I'll give him full credit for this. I've referenced this before, but he says, treat it like your lazy graduate assistant, right? And you'll never be disappointed with what it comes back with because it is oftentimes eighty percent of the way there.

It's a little sloppy. The references aren't great, right? But it does a lot of the work that gets you to the, across the finish line that you wouldn't have to do from scratch, right? And I think when you start looking at it as a collaborative tool like that, you're setting your expectations correctly.

You understand that it is, you're not training your replacement. You are training a tool that is going to work with you for the foreseeable future. One of the things that we've talked about too, is the fact that we almost hit, I don't know if I'd call it a point of diminishing returns, but an interesting, almost plateauing of some of the gee whiz aspects of these tools, right? We got, okay, now it's trained on three trillion parameters or whatever.

And now we're looking at micro and mini versions of these tools that are more fit for tasks, consume less energy, that are cheaper to run, things like that. And we're seeing less of the, you know, they're getting more accurate in some ways, in some ways not, right? What's next? Like, where we go?

Because I think, like you said, it can feel very much like you got left behind, but there's so many resources out there, especially in education, that is so collaborative. And we'll actually, you know, within the show notes of the podcast, we'll add a link to our AI resource hub that has a lot of stuff within the infrastructure community. But we've also got, as well as the study that we mentioned earlier, but it's such a collaborative environment. No one's starting from scratch.

You can borrow somebody else's acceptable use policy. You can borrow somebody else's language for their syllabus for educators. Like, there's We publish these out there because we want other schools to see them and share them and build off of those. So to me, that's one of those Most industries aren't quite that collaborative, and that's one of the benefits of being in education.

What's next? I mean, what do we have to look forward to in the next couple of years?

Oh, man. I thought I don't even know what I'm gonna have for breakfast tomorrow. Yeah. It's a catch up to do. So, you know, I I think you're right.

What we're seeing in the market, it looks like a kind of a local maximum Yeah.

Or a plateau. Yeah. Right? Where model providers, the OpenAI's, the Anthropix, the Googles, training models at the frontier, these big trillion rammer adding rammer Yep.

The massive words. Incredibly expensive. Yeah. And, you know, that's increasingly getting us less and less because these models are so good.

And so I think what you will see is providers first focus a lot on these smaller models. Yep. Trying to to get the same level of performance out of something that is smaller and faster. Now I think that's that's great for education from a cost perspective.

These models are far more affordable to run. It's also great for us as, creatures that breathe oxygen and and then need a habitable planet and drink water because those models are better for the environment too. Right? Cost is a really great proxy for environmental impact models.

So I would keep an eye on what's happening in that smaller models. Yes. Great example of this, you know, OpenAR's Anthropic just released Haiku four point five, which is a model that's performing better than the prior generation at a much lower cost. And thanks to their investments in skills and OpenAI's investments and apps and things like this Yeah.

You're seeing these models able to do more and more. And I think you will see providers focus a lot more at that platform level Yeah. Which should make it easy for us to use these things in more novel and interesting tools.

Well, we've talked a lot about our friends at AWS. I know I spent some time with them last week. You were spending some time with them this afternoon and the rest of week, but like, tell us a little bit more about Bedrock large language model approach that they've got. Really does enable customers, anybody, to use these tools in really novel ways. But I think last time I was like, oh, there's like seven, and you're like, no, no, no. There's like twenty eight large language models.

Even that's moving faster, but talk a little bit more about that.

Yeah. So, I mean, what I like about AWS' approach with Bedrock is that it recognizes that model choice or provider choice is not a one size fits all, even though same institution or in same company.

Yeah.

And so Bedrock gives you, if you're a programmer, have one API and then gives you access to, you know, dozens of models behind that. So if I'm I may say, you know what? I just want something that's gonna summarize discussion threads for me or summarize research papers that I'm gonna read, which is something I use AI for quite a bit Yeah. Personally, I may not need the biggest and the best model for it.

Right? That's kind of a solved problem at this point. So I can using Bedrock, I could pick a smaller model that's gonna save me money, that's more fit for purpose, that's faster, and then I can save the more expensive model for things that really are cutting edge. Yeah.

You know, if I'm if I'm having to go crawl the Internet to go collect a hundred sources on a topic for me, I want something that's a little more discerning and a little more capable.

Yeah.

So I think that type of kind of choose your own adventure is a really powerful component of your technology staff because I think it gives you a lot of freedom in deciding what your AI journey looks like, which again, I cannot say it enough. Your AI journey shouldn't be whatever I tell you it is, and it shouldn't just be use more AI.

Yeah. No. I think that's spot on. And I love that. That's kind of your mantra.

I've heard you say that more than once. So So exciting. Educational, we've got a small overhead announcement happening at the same time. I guess one of the interesting things about Educause is always like how many new technologies are here, how many existing technologies that have been around for a while are here.

Always good to see old friends stopping by the booth is, you know, a really good stuff. But I guess what's, is there an underlying trend? I'm not seeing necessarily like a, we saw proliferation of AI tools last couple of years. I think there's, it's more now some of the old standards. There's some interesting poster sessions I've been going on doing some technologies. But is there anything that stood out as as being really novel from this year's EDUCAUSE?

Yeah. I, you know, I think you're right.

I think a lot of a lot of schools are, again, realizing that their their goals and their problems didn't change a lot Yeah.

For the last couple years.

Yeah.

And so they're really looking for good technology partners more than they're looking for kind of The whiz bang.

The whiz GWiz tools. Yeah.

I mean, there's there's nothing that's gonna just solve all your problems here. Now something I have seen some discussion on this year, which I think is really exciting and gets back to some of our discussion around Workhorse Development is being able to tag content with Oh, yeah.

Skills it represents. Yep. Being able to recognize learner achievement in smaller ways, which, you know, it was a hard step to scale. We've talking about this stuff for a long time, but we just didn't do it because we didn't have enough people or enough money.

Yeah. And I think AI is really opening some awesome doors there Yeah. In a way that I'm I'm seeing some schools now really seriously look at how they can break their existing courses down into micro format Yep. How they can build up badging programs in ways that are are cost effective and scalable.

Yeah. And there's that, that you're sitting on these mountains of content. How do you actually map it in a way that would make it even usable if you wanted to, right? For credentials, for smaller courses, that kind of thing.

And I think that, yeah, that's incredibly interesting. Yeah. I mean, I think that is a good high level overview of what's happening tonight. We've got an event at Post Malos restaurants, which I'm excited for because, you know, he's Utah native.

He's not that far from us. Yeah. We've adopted him as our own, but he has a he has an Nashville bar. And so there's a not a zero percent chance it'll be there.

And, so I'm I'm just gonna assume he's going to be I get to hang out with Post Malone.

I've gotta say, I mean, if, you know, if you're a Utahan, I think you have a responsibility to go to to Post Malone's bar in Nashville. But it's like it's like going to Utah in Tennessee. It's this beautiful thing. Yeah.

Man, I oh, I don't know how you wanna end this. Right? I mean, you know what we do if in my family? Maybe we can have this.

Yeah.

You could I Yeah.

Absolutely. Hosts. Yeah.

Was not bad. Okay. We like to have family dinners, play a little game called, peak and pit, where we say something really great that happened in the day, or we say something that wasn't so great. And I'm curious to hear from you. You know, you've had a lot of conversations here. What are you most excited about in EdTech right now? And maybe you do this in the opposite order so we don't end on a downer, but, you know, is there anything that you're, a little nervous about or something that you feel like we need to as an industry kind of rally around to figure out?

Well, I'll start with the pit piece. I still think we've got to bring educators along. I think it's very institutionally based whether or not there's a culture of innovation and adoption and AI is table stakes. It's not going away.

We're beyond that conversation. Know, there was the, we know there's resistance within, in the market for that and a negative aspect. And to be fair, we're both AI optimists, and it is possible that there are challenges with AI. We've got to be cautious.

We've got to eat our vegetables, as we always say, around data privacy and data security. But there's so much opportunity with these tools. And at this point, it's a little like saying, don't use the internet. It's not good.

Or don't like, we're beyond that. We've got to move forward in a positive direction. And I don't know how we bring all of our peers along in the same way. The other end of that is I'm really excited to start seeing some of this pedagogical use of these tools, right?

Not just, you know, last year we, and a lot of our features are really focused on saving educators time, but because of the open architecture we have, we really create this opportunity to create whatever ecosystem environment you want to at an individual course level even. And so how do we adopt these tools in a way that is creating more personalized experience for students, a more engaging experience for students, a more pleasant experience for educators. Right? So they're seeing the spark of engagement with their students.

Is the reason they got into it in the first place. Right? Like, how do we how do we keep kind of stoking that fire? And I'm starting to see that AI start applied to that.

And that's I that actually gets me really excited.

That's fantastic. Okay. Let's see. So my pit, you know, I so just this morning, OpenAI announced some new researcher in safety.

Yeah. And the good news of the report is that they improved their models' abilities to detect negative conversations. So the conversations of people who may be having, some type of manic episode or, you know, a mental illness challenge. They increased reliability of the model detect these from fifty percent up to eighty and sometimes ninety percent.

Yeah. Which is great.

The bad news there for me is that the numbers they published said that, you know, point one five percent of of all users who use ChatGPT in a week have a conversation like this or a conversation where they have some emotional Which which seems small.

That's a very small number. Point one five percent. What's number gets a little bigger when you realize eight hundred million people a week are using ChatGPT.

So talking twelve million Twelve million legal.

Jeez.

And, you know, I I'm with you. I'm a real optimist. I spend a lot of time thinking about what's what's going well and how are things are going. That was a real reminder to me this morning that I owe a real debt to kind of all of my brothers and sisters on Earth to care for these folks.

You know? And I I so I I guess I applaud model providers that are doing work here, but I think as a planet, we've all gotta chip in and do our part two. Now my peak, candidly, we mentioned environmental impact a little bit. Yeah.

This is something I've been very worried about in the past. I've spoken with a lot of educators, a lot of students that have been very worried about them in the I have never been more optimistic about the environmental future of large language models than I am in today. That's great to hear. You know, there was for a long time, these models were real black box, so that I don't have to know what was going on.

We were all just kinda making guesses as to the environmental impact. I, you know, Google, again, shout out to them. They they published some some research in the last few months about their environmental impact, and it, you know, it wasn't totally comprehensive, but it was pretty good. And it actually showed that we seem to be doing the right things Yeah.

In technology industry. And I it gets me excited because it means, again, that more people are gonna be able to use these tools Yeah. But they'll also have a planet, hopefully, to do it. So that that's what I leave here with.

Yeah.

So I like that.

That is a great wrap up. And so, you know, we'll do another episode at some point here, maybe a recap of the last couple of days. But that is your a wrap up of the first day here at Educators twenty twenty five on the Educators three thousand podcast. Thank you for joining us. Thank you for audience here live in the booth, and we'll see you next time.

Thanks for listening to this episode of Educast three thousand. Don't forget to like, subscribe, and drop us a review on your favorite podcast players so you don't miss an episode. If you have a topic you'd like us to explore more, please email us at InstructureCast at Instructure dot com, or you can drop us a line on any of the socials. You can find more contact info in the show notes. Thanks for listening, and we'll catch you on the episode of Educast three thousand.