Rethinking disruption

00:00 Introduction

Sanjay SrivastavaHello, welcome to the podcast. I am Sanjay Srivastava, the host of The Learning Curve, and I'm also the CEO of Vocareum. My guest today is Michael Horn. In 2008, Michael co-authored a book with Clayton Christensen and Curtis Johnson called Disrupting Class. It took the theory of disruptive innovation, the idea that had already explained why great companies fail, and turned it on something far more consequential: our schools. It became one of the most influential books written about education this century, and it has been argued about ever since. Michael went on to co-found the Clayton Christensen Institute. He teaches at the Harvard Graduate School of Education. He has written eight books, including Blended, Choosing College, From Reopen to Reinvent, and most recently the bestseller Job Moves. He co-hosts Class Disrupted and Future U, and writes The Future of Education. He was in front of Congress just weeks ago testifying on AI in higher education, and hopefully we get to talk a little bit about that testimony. And he has spent 20 years doing something rare: publicly examining whether his own theory worked out. Michael, welcome to The Learning Curve.

Michael HornYou and I have obviously known each other for over a decade at this point. I was reflecting on that when we met at UVA years ago. It's great to see you having become an entrepreneur in this space, not just in others now as well. So, love it.

Sanjay SrivastavaYeah, and I remember that UVA meeting too. Just to give you context on why I was excited about it: I started my first company in 1995, three or four years after graduation, in Silicon Valley, and I decided to bootstrap it. I was an engineer, so I was trying to read everything I could. In those days I remember books like Crossing the Chasm, Good to Great, and Regis McKenna's Relationship Marketing. But 1997, I think, was the year The Innovator's Dilemma hit, and that was a really big deal, especially at a bootstrapped company, because it gave me a language to talk about how to view the world. Where is this disruption going to be sustaining, and where is there an opportunity for me to get in? It gave me a language for unserved markets. So it was a big deal. And what blends in my mind is not only The Innovator's Dilemma but The Innovator's Solution. But when you and Clay wrote Disrupting Class, that was a big deal for me personally, because I was trying to get into the education space and it gave me some language and context I was already familiar with. So it's really awesome to have you here. Just to get going: for the people who don't know, can you go back to Christensen's original thesis in The Innovator's Dilemma and The Innovator's Solution, and then how you applied it in Disrupting Class?

02:40 What disruptive innovation actually claims, and why K–12 stalled

Michael HornYeah, absolutely. And it's interesting: a lot of people say underserved customers need disruptive innovation. You're like, no, no, no. They want more stuff, they want the sustaining innovations. It's truly those that do not have access to existing leading companies' products and services at the time. Disruptive innovations come in with something powered by a new technology that's more primitive, simpler, easier to use, more affordable, and takes root with them. And then it starts to get better and better. And the people who are using the traditional products or services of the market, as it gets better, they start to move out to the disruptive innovation one by one. They're delighted with something that's good enough now, but it's also more affordable, simple, accessible and so forth. And so this is how computing changed, automobiles, retail. You can tell the story in product industries, service industries and the like.

Michael HornAnd what we tried to do in education was figure out — it was clear our public schools in the United States were stuck. With what I would call a factory model, a monolithic model: batching kids by age, delivering content to them, and then sending them out the other side and being surprised that some percentage of them got it, a huge percentage didn't really, and some percentage didn't at all. When in fact, of course, that's how it happens, because learning is extremely jagged. It's not linear. We move at different paces, we have different background experiences. And so the question was, could we use this sort of body of innovation theory to help schools get a little unstuck?

Michael HornAnd what I would tell you is the first thing you see is that, at least in the United States, there are no unserved students. And so we tried to get a little creative and say, okay, well, maybe the disruptive innovation occurs within schools, in how the classroom operates. And there there's lots of non-consumption or unserved, right? There's the advanced classes that schools can't offer, or credit recovery, or things of that nature. And I think what you see is, online learning, digital learning was coming up at the same time. That felt like a viable technology enabler. And it did take off in line with disruptive innovations, but because it was still within the traditional system of schooling, it did not create the personalization and mastery-based learning that we had hoped for and argued for in the book.

Michael HornBecause at the end of the day, the metrics, if you will, that it was judged by were still seat time — that process of, did you cover the materials, send them out, regardless of how much they mastered and so forth. So in many ways, we got a disruptive innovation that got crammed into the schooling model we had argued about. And it hasn't produced great outcomes as a result, I would argue. Clearly digital disrupted analog in terms of delivery, but not the gains that we wanted.

05:33 Education savings accounts, COVID and microschools

Michael HornNow, I will tell you there's another chance in K–12 education, I think, for this maybe to come out differently at the moment. And that's purely because we're seeing a bunch of states offer these education savings accounts, where they'll actually now pay families to choose different schools, different educational models, tutoring, et cetera. And all of a sudden that quote-unquote free public school maybe looks a little less free than it used to, when you're forsaking some numbers of thousands of dollars from the state. We'll see. We don't know yet, but it's possible that there's actually a market forming that could be more in line with Christensen's earlier studies. But that lack of non-consumption, that lack of unserved students, that really was the thing that, more than anything, gummed up the disruption story in the K–12 world.

Sanjay SrivastavaOne of the things, as you're talking about ESAs, which is not a technology but a public policy — but also COVID maybe sort of forced a non-consumption. I know you recently talked to Prenda. I myself went and invested in a company which is around the microschool space, and that seems to be growing. Maybe non-consumption was sort of forced, because the kids were staying at home and parents were saying it was non-consumption.

Michael HornI think that's right. I think it's sort of the combination. The ESAs created a price value that was missing before. And then COVID created a non-consumption, or also dissatisfaction. I think parents for the first time were like, wait, what the heck are you doing with my kids? And some of the screen-time blowback at the moment I also think is part of that, because they were like, wait a minute, how many hours is my kid watching YouTube for? That's not active learning. That's not what I want. I think it's all part of it. And you're right, microschools are growing extremely rapidly at the moment. Is that the lift, I guess? I think it's possible, as long as they're truly low cost in nature. I'm a little more skeptical of some of the microschools that have come in at really high price points, of them somehow turning over the whole market.

Sanjay SrivastavaYou know, that's a really good point. Maybe just the language around personalization, which started even at Khan Academy — I'm not sure of the history — but with Fusion Academy, I send my grandson there and I absolutely love the model. You can do some back-of-the-napkin calculation about the cost, and to me it feels like, while it is not delivered at a low-cost price point right now, potentially it can be. That feels pretty exciting too.

Michael HornYeah, I think so. And for those that don't know, Fusion is sort of an ultra-personalized model, and we have one near us for sure in Massachusetts. I again think part of this is it's not going to be one school model that overturns this, but it's probably a collection. It's like Primer meets Prenda, OpenEd, Fusion. Alpha probably has some role in this. It's like death by a thousand cuts. And I think we could get to a world in which in the future there's maybe a series of scaled low-cost microschools, and then there's your mom-and-pops that are in the community, and then there's some percentage probably still in the public schools, I would imagine. And that's sort of your future of education. But I think it is families saying, just because I live here, I don't have to just take this option. I can assemble this. I think parents are starting to become more inquisitive and active in that process, which will be a good thing in terms of moving this forward.

08:32 Why disruption needs a technology enabler

Sanjay SrivastavaOne of the things which I want the listeners to actually hear about: when I was talking to you over email, you made the point that the disruption theory was not about tech. And I tended to think about tech, because I happen to be in tech, and at Vocareum. For us it was pretty exciting, because Moore's law — a lot of people might not realize this — actually provides a steady pace, an enabling technology, which often does disrupt big stuff. The biggest disruption, of course, is AI. All of a sudden compute became available at a price point where you can do some crazy model training. But you made the point that disruption theory was not built around technology, but on disk drives and steel mills. There's a technology enabler which is needed. Is that a fair way to describe the point you're making?

Michael HornI think a hundred percent. And in some ways, you're embarrassed when you read your old writings. I was reading some old blog I found from years back, where I was like, could community colleges be the disruptive innovation in higher education? And you're like, actually not, because they don't have a technology enabler. So tech is certainly a key part of this, because you have to be able to come in with that primitive offering. And to your point of Moore's law, it's got to improve at a trajectory — but without taking on all the cost and centralization of the thing that you're replacing.

Michael HornAnd so a tech enabler is key. And I think it's that sort of game-theory element of disruptive innovation: if we come in here with these unserved individuals, this group is gonna be thrilled, because they're like, well, we weren't serving them to begin with, and they're gonna be happy to cede that part of the market. And then as they start to go upmarket and intrude, then you have a little asymmetric competition, if you will, that tends to favor the disruptive innovator. The classic story was the mini-mills. When they came in, their profit margins always looked unattractive relative to the integrated mills at the time. And so they were happy to say, you do rebar, you do angle iron, we'll just cede that part of the market consistently and go to the higher value, our best customers. Of course the danger of that is if you keep going upmarket to your best customers, at some point there are very few of those left.

Sanjay SrivastavaRight. And I'm actually refreshing the book again, Disrupting Class. One of the classic ones is DEC — making a lot of money, doing great. They were our customer when I started. But of course we know what happened with the PCs.

Michael HornYep. PCs, and their CEO famously said, who the heck wants a computer in their home? Well, ninety-plus percent of us, it turns out.

11:30 Radio, TV, computers, the internet, now AI

Sanjay SrivastavaWhen I think about technology, since I have been in Silicon Valley, I think about PCs, and then it became mobile, then the internet, then the public cloud, and of course AI now, as the broad, big changes. But as far as education is concerned, it feels like maybe it was first computers, and then online was supposed to be the enabler, and then now it's AI. Is that fair — that those are the vectors to look at, as far as the potential to disrupt education?

Michael HornI think that's right. Although I would just note that if you look back further over history, the chalkboard, I would argue, was the classic sustaining innovation for the classroom. But then when the radio and television came out, you definitely have prognosticators even then saying we won't have to go to school anymore, everyone will learn around their fireplace or whatever it was. So we have seen these predictions before: radio, TV, computer, internet, now AI. And I think the consistent thing that you could say over and over again is, where are the unserved? Where's the real sense that you're being overserved by the existing school system?

12:07 Higher ed, AI, and “it's not a bug, it's the benefit”

Sanjay SrivastavaSo let's talk about higher ed, which feels like a more fertile market, right? Because unlike K–12 there's no mandate that everybody has to go to school. And higher ed is pretty interesting just in the last couple of weeks. I'm talking about AI. I'm sure you saw the Brown professor saying that when you took students from a take-home exam to an in-person exam, the performance dropped. The University of Chicago law school comes out with a policy. Let's start with AI, because I'm really curious about it. The talk is still that AI is at least causing a crisis, maybe not a disruption — people worrying about assessment, which is so embedded in the architecture of what schools do, and people worrying about cognitive offloading. We know what the problem is. Where do you think it's going?

Michael HornI think the biggest issue at the moment, and this is what we got right actually in Disrupting Class, was that these problems of new technology almost always look like threats to the existing organizations, because they upend their processes, their sense of identity, how they do things and so forth. And it really does take new organizations, new business models, new institutions coming along that say, actually, that's not a detraction. That's actually a feature, right? I guess the line would be, it's not a bug, it's the benefit.

Michael HornAnd we haven't had that really yet. Western Governors University, obviously, online competency-based university — we know that they're playing with an AI-native university. We know that UMass Global is launching an AI-native MBA program. Actually, they've launched it. It's $3,000 a kid. I think those places are having very different conversations around this, because they're saying, okay, accept that that's going to happen in a world of zero-sum competition and take-home homework and so forth. What we're interested in is how do you work alongside the AI to actually see what students are really doing, and assess very differently, and think about the work, and sort of create a system in which students feel like they're cheating themselves if they outsource it all to AI.

Michael HornAnd I don't think the traditional academy has made that flip yet. It's still about which one of you is better than the other, which therefore, if that's the race, I'm just gonna use AI to come up with the better-looking performance. It's less of, hey, as a faculty member, I actually want every single one of you to master this, which does require you to do the hard work and figure it out. But the processes and systems have to then therefore look very different inside of an institution. And I frankly think it's going to be a few years until traditional institutions can even really absorb that conversation meaningfully. There's certainly leaders at institutions — Ohio State, Purdue, USC and elsewhere — I think are thinking more like this. But I think it's really hard to turn around these Titanics of operations.

Sanjay SrivastavaYeah, I have empathy for this. Going back to the architecture: as an employer, I worry about two things. A few months ago I used to worry about whether people can use AI effectively or not, and that has changed. Now when I'm thinking about hiring people I have two concerns. One is, are people expert enough in their domain that they know when to use AI and how to use it? And second, of course, I worry about whether they know how to use AI effectively. And if I had to pick one of them, I would say we'll teach about AI, but the whole deep foundational domain knowledge, which you learn through the years of schooling, that is much harder. So I'm in favor of a law school saying that in certain classes you can't use AI. It seems like at least today that's the only answer. But I want to talk about UMass — I'm not familiar with it. What are they doing with the AI-native university? How does it work?

15:51 What an AI-native MBA looks like

Michael HornYeah, first let me just say I couldn't agree more with you, so let me put that out there on the record. I think for novice learners, don't use AI unless it's a teaching tool. And that's how UMass Global is using it, which is, as I understand it, it's one faculty member for the entire MBA program for 50 students, with AI doing most of the teaching and mediating and assessing itself and giving feedback. And my assumption is that they really spent time training it on the curricular materials, so that its answers are limited to some degree by the material, I would imagine. I haven't seen it up close, but I think that's how they're using it. And my guess is that it's dialogic, but I don't know that — that it's a conversation with the students. And you could use an outside AI to do the work, but then when you're having the conversation in real time with the AI, it's gonna pick up either that you don't know it or that you're pausing to use something else. And so I think that's how they've done it, but that's an open question. I'm assuming there's a lot of group work as well around real-world projects and delivery for folks, but I don't know that exactly either.

Michael HornIf I were doing something, I would do very little AI while you're doing that deep foundational knowledge that you described. And then as you were doing projects, I would say, please use the AI to do even deeper work, right? So that I can see you can do the combination of the two.

17:08 Practice, not AI: the UC San Diego math result

Sanjay SrivastavaRight, that totally makes sense. On your point about WGU — I don't completely know, but one thing which was exciting to see is, can you use AI again to go after the unserved market, increase the funnel even more, to people who can be brought in to the curriculum? I don't know that I got an opportunity to mention this to you, but we did a two-year-long project with Vocareum at UCSD, solely focused on developmental math. I'm assuming you have been following this whole crisis in the UCs.

Michael HornYep, on the math. And UCSD was the poster child, yeah.

Sanjay SrivastavaSo UCSD knew it was going to happen, so they brought us in to say, hey, we have been doing some research, but we need to make sure that it gets productized. Can you work with us? And that was pretty exciting. So we finally ended up publishing. The report says that, taking into account all the variables of the students who are coming in and the teacher and all that, the failure rate dropped by two thirds — so 70% fewer students were failing. And the thing was AI, but the discovery for them was that AI did not make the outcome better. What made the outcome better was practice. If you practiced more, you did better, regardless of where you started. But what AI did for some learners is it just reduced the barrier for them to practice. It's pretty fascinating. So we are trying to scale this up.

Michael HornWell, and that makes sense, right? I mean, the line goes, he who does the work does the learning. And there's unfortunately this thing in education against — they call it drill and kill. But I don't know any field in which you can become an expert without practicing. And so you do need your reps at some point.

Sanjay SrivastavaRight, so again, the architecture for this one was: Monday you got in homework with AI, Wednesday the AI got taken away, and Friday you have to do a test without AI. So that sort of solved the problem of saying, hey, if you remove AI, then the performance drops.

Michael HornYou're really gonna do the work to learn this thing. Yeah.

18:56 What has actually disrupted higher ed, and the acquisition-cost problem

Sanjay SrivastavaSo what do you think? Going back to higher ed and the workforce, what's your view of the last 20 years — what disruption has happened in higher ed, and any other vectors like WGU and SNHU that you talked about, getting to people at scale?

Michael HornYeah, I think those have been the clearest disruptors in the higher ed market. They've been sort of classic in some ways. And now with the demographic decline of traditional 18-year-olds, I think we're about to see them really eat up — the combination of all these factors, we're gonna see a lot of colleges closing. It's already started, frankly, but it's gonna accelerate over the next several years. But I think for other institutions, they're gonna have to think about, first, go back to what's our purpose, right? How do we design around that? Really, what you did with UCSD — I think that's a great model. People are gonna have to sort of department by department, domain by domain, go back to those first principles and say, this is not a technology problem, first and foremost. It's a problem of design. And then how do we use the technology to accelerate where appropriate and take it away where it's actually interfering with the actual work itself? And so that's what's required.

Michael HornThe other thing I would say, what's not certain, I think there'll be a lot more entrants coming into the higher ed market powered by AI. The nut you have to crack, it turns out, is not technology. It's student acquisition costs, right? You've got to get it low enough relative to the cost of delivery and the price that you charge. And that's where I've seen most possible disruptions, frankly, fail. I think coding boot camps, when we first met, they were fascinating, got off to a great start. But it turned out if your model was to throw people out in the wild every three months, you had to recruit a lot more to build up over time. And that becomes really hard.

Michael HornAnd so maybe the biggest secret, I would argue, to Western Governors University's success is that, whereas the online learning market has gotten more expensive to acquire students every single year, Western Governors' costs to acquire students have actually gone down every single year. And I think it's because of referral. I think roughly 50% of their students are referred from past graduates of the program, which is actually the best way to hear about a program — someone's saying, this was a great deal, this is what it did for me in my career. So my instinct is any disruptors that come along to really crack the nut, they're gonna probably have to have that similar thing, where you get real endorsers out there in the market helping send students to you at no cost, because you're doing such a good job of delivering value.

Sanjay SrivastavaYeah, and for WGU, it's like my own step kids have gone there, and they are working, and you say, I need to get some classes done so I can take my nursing certification. I mean, there's nothing better, right? It just works.

Michael HornAnd it's a no-brainer, right? For those of us close to it, I think that's the thing. It's almost a purpose brand, right? You're working and you need this. Well, clearly that's gonna be the answer. It's almost like, I'm gonna go Google that. When you become that in an industry, then things become — I won't say easier, but they start to roll downhill a little bit for you.

22:00 Why online programs still recruit locally, and the pull to go global

Sanjay SrivastavaYou make a great point about acquisition costs, and I've seen the same thing. Even the large universities want to launch an online program, and I think there's such a geographical bias, like half the students come from within your geography. So even with really, really strong brands, how do you lower the cost if you're just a new entrant? Really unsolved problem.

Michael HornIt's really hard. And to your point, I think it's sixty percent roughly of students enroll within fifty miles of where they live, even today. And that's true in the online market, right? And so to your point, if Penn State launches an online program, they're gonna get considerable numbers for two reasons. One, the brand name is so big. But two, if you graduate from Penn State, you all of a sudden are part of this alumni base, which largely lives and works around Pennsylvania. And that's an enormous asset, because the job market is still a who-you-know market, and probably even more so now today because of AI. And who you connect with and associate with is a big deal. And so Western Governors, Southern New Hampshire's, Arizona States — those are sort of the exception that have been able to figure out how to go national and build networks. Most people have been stuck, because labor markets are still regional in nature for the most part.

Michael HornI will say that I think there's this other part of it, and you've seen it in other markets, right? Which is, as disruption occurs, it sort of hollows out the middle of the market, and there's a lot of consolidation at the top. And some of those players do go for scale for quite a while. And so I kind of wonder — NYU, Vanderbilt, Arizona State, Northeastern, right? These are becoming worldwide universities with campuses in multiple countries. And I think part of it is that they're realizing that the market has gotten so big globally that they can expand significantly and still be selective, if you will, and so might as well go for bigger market share. But also, if you're Vanderbilt, your market only being Nashville, Tennessee — Nashville's a wonderful city, but it's not going to provide jobs for someone who wants to be in India. And so starting to open up campuses in different countries, cities, et cetera, becomes core to that strategy, is sort of my hypothesis. But it's something I've been really interested in, these brands taking their US position and going global with them. It's really interesting.

Sanjay SrivastavaYeah, is it like Nike? Does it travel? I guess you'll find out, right? I think it does, but we'll see.

24:15 Workforce learning and Reach University's apprenticeship degrees

Sanjay SrivastavaWe didn't talk about workforce learning. Any thoughts on what has disrupted and what do you think is going to happen?

Michael HornYeah, well, you know this market well, right? I think there's been a lot of disruption in terms of traditional trainings that were delivered in person, like the Pluralsights of the world, or frankly YouTube, and all these tools, or even GitHub, that have allowed technical people in particular, but not exclusively, to constantly be reskilling and upskilling in their jobs and bring a lot of people in. I think it's hard because there wasn't an obvious at-scale player before delivering that sort of workforce learning.

Michael HornThe thing that we're not sure about yet is, will there be some sort of apprenticeship model that goes global, right? Multiverse looked like it might be. But then here comes Reach University, new university, with apprenticeship degrees in teaching and now allied health professions and healthcare, and they're growing significantly fast. I mean, there's several thousand students already. You know, could that be something that sort of sits in between higher ed and workforce entry? I think that's something to watch.

Michael HornI guess the last thing I would say is, in some ways, AI-created simulations, just frankly having ChatGPT alongside you as you're doing your work as a tutor, those may upend a little bit what we think of around all this. And so that's something to pay attention to as well. But certainly there have been a lot of companies that have come into that space and become pretty big pretty quickly.

Sanjay SrivastavaYes, you're talking about Reach. I know a lot of people, but even I was not that familiar with it until pretty recently. That sounds really interesting. Do you want to describe it for people who don't know? I mean, I know you are sort of involved with them, right?

Michael HornLoosely — I'm not officially in any way, I'm just friendly with them. But I find them fascinating, probably the same way you do. They basically say they started with schools, and rural schools in particular, that had shortages of teachers. And they said, well, you have paraprofessionals and people who probably work in the lunch line or whatever else who are really good with kids and show promise, but they can't leave. They can't quit and enroll in a teacher's college to get their teaching license, because it's too expensive. And even online school or night school, maybe, it's like an eight-year proposition. That's not gonna work.

Michael HornBut they said, what if we sort of flip the model, where when they're actually employed by the school district — and you could pick the ones that have the most promise that you want to enroll in this program — you keep paying them as an employee, like an apprentice. But now their work, you start to put them in different settings that would correspond to the courses you'd have to take for teacher education. And at night, they'll hop online for an hour and a half to do the theoretical basis of teaching with a cohort. And in that way they'll earn their teaching degree. So it's like the homework, if you will, is the employed time. It's the practicum time, when you're in a classroom, and you're getting the online learning for the liberal arts and the theory basis of education.

Michael HornAnd as a result, they've completely made the cost of this essentially free to the students. No loans required. And they've taken away the opportunity cost, because you're able to continue to work while you're getting that degree, and your work experience counts for credit effectively, as the homework part of your degree. And now they're moving into healthcare, as I said. And so I agree, it's a fascinating model. I'm gonna get the numbers wrong, but I wanna say it's like 4,000 students or something like that. And so, Western Governors is the biggest teacher preparation program in the country; they're probably second or third already. So it's quite interesting, I agree.

Sanjay SrivastavaYeah, that sounds interesting. And Vocareum was started with this called vocation in the name, because that's the model I was really hoping for — can you get paid to actually learn? Never quite got there. But I'm glad to see that someone is actually being able to pull it off.

Michael HornI remember. I mentioned the Pluralsights and the Simplilearns and folks like that, that have gone one path. I think what's so different about Reach is that they are still playing through the accreditation system, and there's a ton of jobs out there that are gated by licensure requirements at the state level. We could debate whether those are good requirements. I think in many cases — I just learned florists have to be licensed in Louisiana, and I'm not sure that's a good use of anyone's time. But as a result, you have to sort of navigate this policy and regulatory environment to even be able to get in the game, to get students for the credentials. And Reach was willing and able to do that, but I think that's hard. And that's been a very high barrier to entry that I hope comes down in the years ahead, so that many more people like you could get their start, but actually acquire students much more easily than has been the case.

Sanjay SrivastavaYeah, I agree with the accreditation. So just going back to my grandson, we had a choice, and the fact that Fusion was saying that they were accredited did actually take a lot of risk off when we were considering it. So I see the point.

Michael HornYeah. And I wish it didn't, because I'm not sure of the value, but it is the way the market works. People are suspicious when they don't see it.

29:09 Predictions: closures, accreditors and experiential learning

Sanjay SrivastavaSo Michael, getting toward the end of the podcast. Anything you want to predict about any disruption in the future, or any final thoughts for listeners?

Michael HornI guess I'll say a couple of things. Clay and I famously predicted in 2012 that a quarter of colleges would close or merge. We're already at over fifteen percent since then. I think it's gonna probably be over twenty-five percent when you fast forward another ten years from now. We've taken a deep dive lately with Steve Schulman looking at the finances, and that'll be a big thing.

Michael HornThe second thing that's gonna happen is at least one, maybe a couple more accreditors will get approved by the Department of Education in the next couple of years. If that happens, then it could lower the barriers to entry in higher education, which could increase disruptive innovation there. I think that would be a great thing.

Michael HornAnd then I guess the third thing I would say is, if there's more states that jump into education savings accounts and they don't put up requirements like accreditation, and they allow families to decide how to use the dollars — they're much looser on that — I think we'll see potential for more growth and disruptive innovation there as well.

Michael HornAnd then the very last thing, this workforce learning piece. I think the biggest thing that's going to be important in the years ahead is, as AI remakes what work looks like, is much more experiential learning. I fully agree with you, you've got to build domain expertise to be able to do this work. But in some cases we're not going to know what that domain expertise really looks like. And so deep real-world experience through projects, simulations and stuff like that is going to become much more valued. The only place I think you're going to be able to get that experience is through education institutions. Because if I'm an employer, I don't know that I want to provide it. And so I'm going to be looking to the schools around me that do really good partnerships and experiential learning opportunities.

Sanjay SrivastavaThank you, Michael. This was great talking to you and really appreciate your time.

Michael HornI appreciate everything you do. Thanks so much, Sanjay.