ASU+GSV 2026 Summit | Monday, April 13, 2026, 11:00 am-11:50 am | Sponsored Partner Programming
Speakers
- Adeel Khan, MagicSchool AI
- Connor Zwick, Speak
- David Rogier, MasterClass
- Larry Chu, Goodwin
Key Takeaways
- A provocative lunchtime panel moderated by Larry Chu (Goodwin) pitted three EdTech leaders against each other on what's worth saving and what's worth burning down in traditional education.
- Adeel Khan (MagicSchool AI) defended the centrality of teacher-student relationships and argued AI should amplify teachers rather than replace them, noting that consumer AI incentives (time-on-platform) don't belong in schools.
- Connor Zwick (Speak) drew a sharp distinction between "learning" and "classroom" as two separate concepts, arguing that AI tutoring delivers 3-5x faster outcomes in language learning and that testing as a concept will become extinct.
- David Rogier (MasterClass) was the most disruptive voice, claiming a 2-year MBA can now be delivered for $100 and that non-top-10 MBA programs won't exist in 5-10 years, citing the MasterClass/University of Chicago Booth/OpenAI 12-week program that drew 12,000 applicants for 500 slots.
- The panel surfaced a key tension: Khan's insistence that "there's an alchemy of wisdom and relationship" in great teaching versus Rogier's blunt observation that "the only reason classrooms exist is cost structure, and that has now changed."
Notable Quotes
"There's an alchemy of wisdom and relationship that creates incredible learning that we should hold on to."
— Adeel Khan (MagicSchool AI)
"Learning and classroom are two different concepts. They've been synonymous because they've been linked for 100 years. But learning will be something you do primarily with an AI tutor in five years."
— Connor Zwick (Speak)
"You can now provide the exact same education for about $100. That's $80,000 a year for four years versus $100. If schools aren't dramatically going to change how they teach, you are dinosaurs."
— David Rogier (MasterClass)
"AI has a perfect memory of everything you've ever said into the app, exactly what you know and don't know. We see people get to fluency at least 3 to 5x faster than traditional methods."
— Connor Zwick (Speak)
"Some of the companies behind these models are the same companies that have been extraordinary at addicting us to social media. Those incentives don't belong in schools."
— Adeel Khan (MagicSchool AI)
Full Transcript
It's incredible to be here. This is my first time at ASU GSV and I must say I'm incredibly impressed with the collection of people, the thought leaders in the room. They gave us a very controversial panel during lunch and the topic is burning down traditional learning. Here comes the disruptors.
We couldn't have a better collection of disruptors today. I've had the privilege of having an almost three decade long career at the forefront of technology and what's happening. And I can tell you that right now we're living through an age of unprecedented speed and change. And so for a while now, I think people have questioned the veracity of our current educational rails, our current educational architecture.
And I'd love to start off with asking the panel what they think in 2026 given what is happening in technology and in AI, what do you think is worth burning down and what do you think is worth saving? Who wants to kick it off? I'll kick it off. I think, first I'll start with what's worth saving.
Before founding Magic School, I was a teacher and a principal and I founded a public high school. And when I hire teachers and they walk through the door, and our first trainings with those teachers, I said that the foundation of our school is gonna be the relationships we keep with our families, the relationships we keep with our kids. And everything else is built on top of it. So certainly we'll have a rigorous curriculum.
Certainly we'll differentiate instruction for students who have different learning styles. But the foundation of learning is a great relationship, and that's worth holding on to in a world where we're spending more time on devices. Our apps that we use every day are designed to keep us on them. So human collaboration, human relationship, that is worth saving.
In terms of what's worth blowing up or getting rid of, I think that believing that there is someone who knows better than the classroom teacher on how to educate the kids in front of them. The person who spends the most time in the educational process with the child, to remove their expertise or believe that that relationship and that daily experience with their learning style, what they understand and what they don't, to believe that there is an external provider that knows that better than the teacher, that's something that we need to throw away. That's present today, and will likely continue to be even noisier as generative AI comes into the world. I'm a big believer in we need to amplify the abilities of teachers to serve students and use AI as a really powerful tool to assist them and to help augment their ability to serve the diverse learners in their classroom.
But there's an alchemy of wisdom and relationship that creates incredible learning that we should hold on to. So Connor and David, so a deal you're building within existing Rails, Connor and David, you're basically saying there's a better other way. And with Speak, Connor, you've basically shown us there's a way to learn languages outside of the classroom. Do you think the traditional sort of seat-based model is broken and gone?
And then David, I'll tee up your question as well. You've brought Hollywood production quality to master class and to learning. Do you believe that traditional lecture is dead, and do you think that traditional learning institutions need to elevate their game? Yes.
Okay, we'll hold, we'll park the yes for a minute, we'll go to Connor first. I think David and I are going to agree on a lot of things here. Yeah, I think to your question around what we see at Speak, so for context, we're trying to build the world's first truly superhuman tutor, starting with language learning. And the traditional way of learning languages, like most things, is really all about, it's centered around number of hours learned.
How many hours have you practiced? And that determines how good you are. And then the second thing we have are assessments. Which, again, are kind of famous for testing a set of skills that are completely different than the set of skills that you're trying to learn, and so you get people gaming tests, etc.
I think this is going to be completely blown up, and we're seeing that with Speak. So we're very focused at Speak on how do we actually build a real measurement of the actual skill that you're trying to master, which is the holy grail of mastery-based learning. And I think the really exciting thing about AI and the technology is that it is possible to do this now. And we're seeing that in the lens of language learning right now, where we can determine in a superhuman way, because AI has a perfect memory of everything you've ever said into the app, exactly what you know and don't know, what you've been able to demonstrate, what you haven't been able to demonstrate.
And then we can completely adjust the lessons accordingly. And as a result, we see people get to their outcome, whether that's full fluency, partial fluency, whatever, much faster, at least 3 to 5x faster than traditional methods. So let me just interject for a minute. How do you reconcile Adil's comment around the human still needs to be in the loop?
There's a very important aspect of having humans in the loop. It sounds like you feel like you can do everything without humans. I think that there are two very different things that are, so far we've combined them as one concept, but they're two concepts. One concept, I think, is learning, and another concept is classroom.
And those are synonymous, because they've been intricately linked for the last 100 years. But learning, I think, will be something that you do primarily for most skills with an AI tutor in five years from now. Because the tutor will be hyper-personalized to you, and it will be, frankly, irresponsible not to have a lot of learning happening one-on-one with a tutor. But I think the classroom is something that we created 100 years ago, during the Industrial Revolution, to have mass childcare, so we could increase the labor pool.
And it's kind of linked with learning, but it's actually serving a slightly different purpose. And I think over time, it'll be interesting to see how these two things co-evolve. But I suspect that we will dedicate part of the time in the classroom for learning, but there will probably be a lot of time, I would guess, where we'll do the other reasons why we have children in school, which is socialization, connection with the teachers. There's a lot of skills that I don't think AI will be better at teaching, and I suspect that that's the way the world will work.
And we'll see that these two concepts are actually a little bit more divorced than we realize. Awesome, David, you jumped in there, and it was a quick yes. You basically created the category of edutainment. Expand on that a little bit, and reflect on how the alternative to traditional learning is challenged by the structured master class.
We know from all the academic work out there that the best way for anybody to learn is one-on-one instruction. So the question has to be asked, why for the past 100 years or 1,000 years do we learn in a classroom of 300 people? Let me ask you, how many of you in undergrad had at least one class that was at least 300 folks in a class? Okay, why?
It's simply cost. It simply was too expensive to get you that instructor one-on-one. That is the only reason why. And we paid for that $80,000 a year for four years.
You can now provide the exact same education for about $100. That's 80,000 a year for four years versus $100. If schools, if L&D departments aren't dramatically gonna change how they teach, you are dinosaurs. The world is completely going to change.
Now, I think in classrooms have done a lot of amazing things, but that is the only reason it exists is because of cost structure, and that now has changed. Adeel, to your point, we know also from all the evidence and work that one-on-one instruction and coaching is really powerful. So if you can add in a person who's gonna coach you, then that's awesome. But I think the day, if you stay in the model we're in today, you're gonna be standing in a lecture hall that is empty.
So how do we get there? That's the question, right? With disruption, sometimes you have gaps. You don't want children who have traditionally only been able to rely on traditional rails of learning to be left behind.
How do we elevate everyone's game? I don't think the technology's quite there yet. I mean, we're trying to build this for language learning, and we're not a superhuman teacher in many ways still. And I don't think the technology's there.
So I think that it's a little bit also confusing to, that can be confused with this idea, well, it's not actually happening. But it is happening, it's just happening in smaller ways right now. And I think over the next few years, it will become dramatically better. And as that happens, I think it will be, as David just said, almost irresponsible if you're an L&D manager.
manager, or you're a dean at a college, to not be thinking about dramatically changing the way that education happens.
Adil, you have a comment on that? Yeah, I mean, I feel like I just want to name, like, I'm not a defender of the status quo here. Like, you know, the status quo has not resulted in extraordinary outcomes for kids, period. We need to be thinking differently about how kids are learning in schools, period.
My school had great results. Relationships with the foundation of the school, that's my formative experience. It matters. And at the same time, like, we need to reimagine what school looks like and what school is, especially in an environment where technology is changing so dramatically.
You know, how we get there, I think, is rethinking what the valuable skills are in the world. I've always thought that, like, leadership's a really important skill. And what is leadership? Leadership is arranging resources in a way to accomplish a really big goal.
Usually, great leaders start with why. Simon said it's a great book, right? They help motivate their teams by letting them know, hey, we're building towards something really, really meaningful. I'm helping you see the big picture of what we're doing, our mission that we're working on.
Cultivating those things in school by doing things like project-based learning, really, really powerful. Equipping students with those tools for generative AI to manifest something in the real world builder experience is really, really powerful. But I still, I think, I'm really skeptical of the idea that, like, me and a device are going to learn as well as me with a team, and someone motivating me, and helping me understand the why behind my work, and those kinds of things. And I challenge the folks on stage of, like, yeah, hey, you know, your most formative learning experience is the engineers that you work with.
They want to know why they're building the thing they're building, don't they? They want to understand that this is going to result in something that's going to make impact in the world. And that's a lot of what a teacher does. They're the leader of the classroom.
They help students understand, hey, you're not just doing that math equation to get it done. Great teachers inspire the why behind that work. They make it relevant to their students, and they create a much more rich learning experience than simply acquiring a skill. I agree with everything you said, except I don't think everybody gets access to a Neil.
If I had you one-on-one, that'd be amazing. If I had you two-on-one, it was you and two-on-one, probably still good. At some ratio, an AI bot that's trained for me is going to be, if it's one at 100, I don't know what it is. But also, Adil, I had instructors and teachers that were pretty bad.
I would have spent less money in therapy if I had an AI bot for second grade than if I had my second grade teacher. I think people are going to need therapy for AI bots as well. That's probably true. That's probably true.
Adil, I want to double-click on one thing. You are deploying within educational institutions. You're helping teachers save time and friction and really elevating their purpose. What barriers have you encountered in trying to get Magic School deployed in school boards and with teachers?
I mean, there's a lot of fear around generative AI usage, both from teachers and especially in usage with students. You can pull up the news and see really scary things that generative AI is doing when it comes to building companionship-type relationships with not just children, but adults. I'm certain that there are model companies that have stories about those companionship relationships that they don't want the world to know because of how scary they are, that we've seen a glimpse of this in the news. But it's important to remind ourselves that some of the companies behind these models are the same companies that have been extraordinary at addicting us to social media.
They understand that their internal metrics at these companies are time on platform. They're incentivized to make sure you are spending more time on the platform. And that is true with generative AI as well. And those incentives don't belong in schools.
So when we win these, we are direct about, with our school districts, that we are here for an educational purpose. Generative AI can have an incredible impact on your students for some of the reasons that have already been shared on stage. But we should be really skeptical around bringing consumer technology into school buildings and expecting that's going to be a solution for the challenges that schools face every day. I could keep on going with this debate, but I'm going to let the- Switching gears for a moment, David, you've spoken a lot about how diploma-based institutions are being disrupted, in particular by AI, and that your focus is on skills-based learning.
How do we increase that methodology of learning and certify people have the right skills that they need? I have a strong point of view on this, so it is a hot take. So I apologize for anybody that this offends. Let's look at MBA programs.
If you aren't in the top 10 MBA program in the country, I'd argue you aren't going to exist the next 5 to 10 years. Here's why. In all the research that we have done, employers now rank MBAs as below a 12-week AI in intensive course. MBAs below that.
Now, I paid $200,000 for an MBA, so I got screwed there, too. We also see on the candidate side that they don't have two years and hundreds of thousands to actually spend on it. If we think about it, I think about my mom and dad, who were both attorneys. They went to law school, and every year they took about a day of a class.
Sorry to, yeah. More than that, actually. More than a day. A couple days a year.
They went to take a class, and they were still up to speed. Now, all that's impossible. The rate of change is happening so much. If you aren't involved in your field, in six months you're probably way out.
You are now a dinosaur. And so with the rate of change happening so fast, I don't think a two-year MBA program works. And so we decided to launch something. It was our first time trying it.
So we launched it with the University of Chicago Booth and OpenAI. And it is equivalent of a 12-week business program that's taught both in person and with AI. In the first couple of weeks, for 500 slots, we've had over 12,000 people apply. And admissions are still going on for another month.
We will be tougher to get into than Harvard Business School. I don't know what else the market needs to see that the signal of schools is just dying. Connor, you also talk a little bit about skills-based, personalized learning. What's your hot take on this?
What's my what? Hot take? Yeah, what's your hot take on skills increasing in terms of the priority stack in learning? Yeah, I mean, I think this is the, I talked about this a little bit earlier, but I think this is the most important part of mastery-based learning.
The entire idea is that around mastery-based learning is like, let's make sure that you've mastered one skill before we go on to the next one. And I think historically, even in a classroom of 30 students, we've had really poor testing or signals to use to understand if people were actually able to achieve the mastery of that skill before moving on to the next one. I think this is a clear area where an AI should be much, much better. And that's what we see with Speak.
And so over the next five years, we believe that testing as a concept will feel extremely antiquated. And we will mostly have, I think, replaced that with something better because the value is extremely high. No matter what the test is, if you can have something where you're reliably, as an organization, whether you're a college trying to admit people and you're trying to understand what they know and where they spike, or you're a company and you're trying to understand the skills of your workforce, the thing that we're working on at Speak is, how do we actually connect directly and accurately to the skills that you need as an L&D manager, as an example, and being able to very accurately predict whether or not those skills are skills that you actually have. So we see testing as a concept being something that goes completely extinct in the years to come.
So that's a good segue into moving more towards the future of what disruption looks like. And I'm going to ask a question that we didn't really talk about before. But Adil, you keyed this off earlier. How do your engineering teams, as all three of you are technologists, Adil, you used to be an educator.
So you wear both hats. How do your engineering teams work with educators to ensure that you are building towards the future and you're building towards what?
what educators who have decades and decades of experience know that we need, think that the model needs to move to. I think that one of the things I agree strongly with that was stated by Connor is this idea of mastery-based learning and that talk about something that we need to leave in the past is classroom-based teaching that teaches to the middle. Being in a classroom where you're way ahead of your peers, you're incredibly bored.
If you're way behind, you're not being served. And AI can be an incredible tool in a classroom like that to help students get accelerated if they're ahead of the curb, supported if they're behind. So I get really excited about using generative AI and novel ways to solve some of these decades-long problems of having 30 kids in a room who all are not the same style learners. There are these persistent edge cases for students, and there's only one teacher who can reach them all.
I think there's some agreement here of where AI can play a really, really meaningful part in that story. And then in terms of your question on how we work with educators, we always start with what their problems are. What are the problems that you're facing in your schooling environment? What are the things that we can do to wrap around supports for you?
Where would AI fit into the challenges that you're having? For us, as we partner directly with school districts, that is our buyer. We certainly build features that teachers love, but really, the business is built through a conversation with the district who says, this is our strategic plan. Where do you fit into this strategic plan?
Because we believe deeply for our local context, this is our best thinking on what's gonna help us drive outcomes for our kids. So it always starts with what are your problems first, and then we come in with, hey, this is where we think generative AI can be really helpful. And we try to push, right? We try to say, hey, you might not be ready for this thing, but we think that it will be such an accelerant to your plan that you should go for it.
But at the same time, we respect the local districts that we work with and their vision for what they want for AI for their children. And that is, it currently looks quite different. I'll say that pretty, like we have schools and districts who partner with us and say, we are not ready to use this with students yet. No matter how much you've assured us that it's safe and you've done all this guardrailed work, we want you to turn that off in the platform.
We actually give them a toggle to turn that off. And then we have districts who are saying, give us more for students. They want more of the stuff that we're hearing on stage. They want more of like self-driven learning, right?
So there's a spectrum of the way that schools and districts in communities are thinking about technology. And that's not even going into like the headlines around generative AI. If I go to the Northeast, and I'm visiting school districts there, they're asking me about climate change. And how are you guys dealing with generative AI and its water use?
Like if I go to the Southeast, it's more of a relational question, it's different. But like generative AI is more than just a technology. It can intersect with your political views. It can intersect with your fears and hopes for the future.
Yeah, I would add. I think for us, we have a team on staff. It's one of our biggest teams of some of the best language teachers, linguists, et cetera, from around the world that help us form our pedagogy. And I think we have also a very intentional thing you mentioned, teachers and classrooms.
I think an important note though from our strategy is from the very beginning for Speak, we identified language learning as a market to go after first because it was a market where we could go directly after adult learners because the mission of the company is to build the world's first superhuman teacher. So we need to move as fast as possible. And we worried that if we were going into classrooms, we would have to worry about all of these other things and it would encumber us, which I think there's a huge opportunity there. But we're really focused on just proving that AI can be a really effective teacher.
And so I think that's been extremely liberating for us because it allows us to really be unconstrained with just trying to build what we think will be the most effective solution and work with learning designers to make that happen. I think most of the pedagogy that we are taught and taught by is outdated and wrong. I would agree with that. I mean, I'll give you an example.
How many of you, I'm in this group, were asked at some point in school, are you a visual learner? Are you a kinesthetic learner, auditory learner? How many of you were asked at some point? There is no evidence that that is true whatsoever.
There's any distinction between any of those things. And if you look at the pedagogy that most that we are taught with, it's an outdated one and most of it isn't true. But then most of it was designed for in the classroom. And now that you can learn much faster with an AI, the pedagogy research is gonna dramatically change.
So I think it's an exciting time, but I think we're starting at almost zero. So we're just out of time, but I wanna do a really quick prediction lightning round. So I imagine it's 2030, we're back at ASU GSV. What should the investors and the educators in this room invest their time or their money in?
What's the one thing, and don't just say AI, obviously, but double-click real quick. Adil, Connor, David? Project-based learning, give kids authentic work, things that they can show off in the world, collaborative experiences with others, build human connection. I tech has traditionally been a really bad industry to invest in.
I do think education will be one of the biggest ways in which AI disrupts the world. So I think it's an interesting area to invest in going forward. I ask for the next five years to build your own teacher. Awesome, thank you to these thought leaders.
Thanks for spending time with us during lunch. Thank you.
This transcript was put together by our friend Philippos Savvides from Arizona State University. The original transcript and additional summit resources are available on GitHub. Licensed under CC BY 4.0.