2-Sigma in 2 Hours: How Alpha Schools are Using AI to Revolutionize Education

Hello, and welcome back to The Cognitive Revolution. Today, I’m speaking with MacKenzie Price, founder of Alpha School and Two-Hour Learning, and pioneer of a new educational model that’s using AI to scale one-on-one, mastery-based tutoring to help students learn 2 to 10 times faster. This allows them to finish their traditional academic subjects in the morning and then spend their afternoons taking field trips, working on independent projects, and otherwise exploring and cultivating their interests.

Importantly, just because the Alpha School model is futuristic and gives kids lots of free time to explore does not mean they’ve gone soft on academic standards. On the contrary, MacKenzie frames learning time as productive struggle. Alpha Schools demand focused effort and expect students to master all of the usual Common Core subjects.

The results speak for themselves. Despite spending just 2 hours per day on academics—which means spending some 20 to 30 hours per subject per year, or just 10 to 15% of the 200 hours per subject per year that’s standard today—Alpha School students learn 2.3 times more per school year than statistical models predict they would in a normal school. Their school-level results on standardized tests are nearly always in the 99th percentile.

In this conversation, we explore both the AI product implementation and the way in which the rest of the school experience is changing. We start with why MacKenzie does not recommend that schools give students access to chatbots and what they do instead, which is to combine an ever-evolving mix of adaptive learning apps—which they both build and buy—with systems that analyze learning rates and patterns and make personalized curriculum recommendations, plus more and more lesson-level personalization and multimodal support all the time.

We also discuss what kids are doing in the afternoon once their academics are complete, how Alpha School is beginning to teach them to use AI as a superpower, and how the Alpha School model can be adapted to different price points, including the possibility of public charter schools and the new homeschool option they’re starting to roll out.

Perhaps most importantly, we explore the evolution of the role of the teacher, which Alpha Schools call guides or coaches, and what it means both for kids and the future of work more generally. As MacKenzie explains, because Alpha’s guides don’t have to worry about delivering academic content, they can focus entirely on motivation, mentorship, and emotional support. The school can both hire for those skills and hold guides accountable for making sure that each and every kid truly loves school.

I find this vision super exciting, both because this does seem like one major area of the economy that AI can positively transform without also disrupting the labor market, and because when I think back on my own education, it was clearly the teachers who made the extra effort to engage me with special projects, one-on-one conversations, and possibilities for my future that I’d never considered who made the biggest impact on me.

I was privileged to have quite a few who made their mark: Mr. Francis and Mr. Rogers in 6th grade; Mrs. McBain for biology; Mr. Vance and Miss Kohut for high school English; Mrs. Voss for AP history; Mr. Andreasco for chemistry and physics; and Mr. Madorski for government. The idea that every kid could have that kind of transformative relationship with teachers, not just as the exception but as the rule, could easily be more impactful than the AI-powered instruction itself.

The bottom line for me, as a parent of a 6-year-old who just finished Montessori kindergarten and a 4-year-old who starts in the fall, is simply that I’m thrilled that the AI revolution is coming to education before my kids have had to endure much time in the one-size-fits-all, lecture-based classroom model that’s dominated education since the Industrial Revolution.

I really look forward to seeing how the combination of AI instruction and human mentorship helps them learn and develop at a rate and in ways that our generation could only have dreamed of. This, it seems to me, is one of the most exciting upsides of the cognitive revolution.

With that, I hope you enjoy this fascinating exploration of the AI revolution in education with MacKenzie Price, founder of Alpha School and Two-Hour Learning.

MacKenzie Price, co-founder of Alpha School and Two-Hour Learning, welcome to The Cognitive Revolution.

Thank you, Nathan. I’m so excited to be here. I love anything cognitive, and revolutions are great, especially when it comes to education. So I’m in.

Yeah. The AI revolution is coming to everything, and education, as you’re demonstrating, is no exception.

And so needed, right? It’s so needed. It’s time for education change. So, yeah, this is great.

Yeah, no doubt. I’m excited for multiple reasons. I’m always interested in the intersection of AI and anything, and regular listeners will know I’m also a parent. It’s the first day of summer vacation here. My 6-year-old just finished his Montessori kindergarten, and my 4-year-old is excited because the next time his big brother goes to school, he gets to go, too.

He’s going to school. Very fun.

Now, I’m mostly going to be worried about the 2-year-old, who’s going to be home by himself, and that’s going to be quite an adjustment come fall. But these are the seasons in life.

You know, one of the things I’ve been saying, Nathan, is that there has never been a more exciting time to be a 5-year-old, especially with what’s happening in education and how much we’re going to see it change in the next 10 years. Your kids are so lucky.

My daughters are 17 and 19, and I so wish that I could shrink them back and start their education journey over, because there are just so many exciting things that are going to happen. I think we’re finally going to really see that change that we’ve needed to see in education for the last couple hundred years.

Yeah, I share that sentiment. It seems like a tough time to be somebody who’s in college, graduating from college, or just entering the job market. But the world—assuming everything goes reasonably well, which is not something I take for granted, but I’m putting the positive hat on for this conversation at least—I think in education, it likely will go very well. It’s other things I’m more concerned about.

It totally is. Well, you think about higher education and the state of higher education. I’ve had a lot of university presidents that I’ve been in conversation with in the last couple of months who are kind of realizing, “What do we do now that the half-life of knowledge is, what, 6 months?” In the age of AI, what is it that we’re doing to equip these people?

I’ve got a daughter who’s in college herself, and we had the conversation this last weekend around, if she’s interested in law or consulting or whatever she’s doing, she’s got to be AI-first. That is the key to making sure that our job skills are successful and translate. You’ve got to figure out how to use AI to give yourself superpowers.

That’s what we actually teach our students, literally starting in kindergarten. We teach our kids how they can use AI tools to give them superpowers and be able to focus on what I always consider the gray frontier. AI is really good at knowing the black and white of existing knowledge, and where humans are going to need to be successful is out on the frontier, pioneering new knowledge and new ways of thinking. That’s what we’re trying to equip our young people to go do.

Perfect. Well, let’s take it from the top. I’ve often quipped, but I don’t know a lot about this, that AI could enable the two-sigma-for-all era in education. Can you unpack that two-sigma effect and give us a state of current understanding of what really matters and what really drives outcomes in education? Just set the stage for what you’re building toward that goal.

Absolutely. I’m going to start by answering that question by going way back. I’m going to go back like 1,000 years to think about Socrates, who was tutoring Plato, who was tutoring Aristotle, who tutored Alexander the Great, who took over most of the known world by the time he was 22 years old.

Back then, we had a one-to-one, mastery-based tutoring system. Of course, it was only saved for the very, very elite—for royalty and the intellectual elite of that time. But that was kind of this rich environment. We think about the Socratic method of discussion and how people could go deep into a topic and become experts.

Then, a couple hundred years ago, with the start of the Industrial Revolution, we had to figure out how to educate the masses and how to do it inexpensively. That was really the birth of the teacher-in-front-of-the-classroom model, and that’s what we’ve been doing for the last couple hundred years.

Now, for the last 40 years, since I was an elementary school student, there have been learning-science research papers that talk about how kids can learn 2, 5, or 10 times faster. Every single one of those papers refers to that being possible in a one-to-one, mastery-based model.

When you think about the two-sigma effect, Bloom had his 2 Sigma Problem, which basically showed you could get a two-sigma improvement in learning when you had this one-to-one, mastery-based model. Every single one of those research papers starts or ends with, “Unfortunately, these types of results are not possible in the teacher-in-front-of-the-classroom, time-based model where every kid goes at the same pace.”

A teacher takes a group of 5th graders who are wildly different levels of knowledge and, starting in August and ending in May, has to figure out how to teach 5th-grade curriculum to kids who already know the material and sit there bored and disengaged, and others who don’t even know their 4th-, 3rd-, or 2nd-grade material and are lost.

And so that's fundamentally what we're looking at: how do you incorporate these learning science principles into a classroom? It hasn't been possible, but that's what's incredible about artificial intelligence. Now that AI has come out, it is finally the tool that's going to allow us to provide precise and specific measurement of not only what a student knows and what a student doesn't know, but also to create personalized lesson plans that allow us to meet every single kid in that one-to-one tutoring experience, where they have a true expert who's able to go with them at the pace and level of education they need.

We're also able to do it to mastery, so that a kid doesn't move forward until they have full mastery and knowledge of that. When you think about when the microscope was invented and what that did for biology, or the telescope and what that did for physics, we have such an awesome opportunity now. Artificial intelligence is going to be the tool that allows us to take learning science and turn it from this sort of fuzzy, unclear thing—where it depends on the teacher, the kid, the environment, what's going on at home, and all these other things—into a precise science with clear inputs and clear outputs.

And that's what is so exciting. Add to that the fact that it's scalable and that it can be accessible for any and all kids, no matter where they come to it. That's what gets me out of bed every day. This is finally the chance that we have to truly bring education and great outcomes to all students.

I love that you brought up the Alexander the Great thing. I was going to bring it up if you didn't, actually. I wondered if you could unpack a little bit what exactly two sigma means. I understand that it's sort of two standard deviations, right, as the measure. So we're, in a sense, saying that we could take the average student and bring their performance up to—if my recollection is right—that would be basically the top 5% of students.

Yeah, you can get really phenomenal results. It's funny. I'll just give you a little bit of my history. I was pretty good in school, but I actually hated school growing up. I was always that kid who would raise my hand and be like, “I'm sorry. Why do I need to know this? I promise you, I'm not going to be doing anything that requires me to understand major details of chemistry or physics or calculus or whatever.”

So it's funny because even my math is a little bit shaky when we look at this. But here's fundamentally what we're looking at: when we put kids in a one-to-one environment, where we're going at the pace they need and providing the level of material they need, that's where you can see these phenomenal results.

Bloom's two-sigma problem is one that a lot of people who get into learning research and stuff pick apart and say, “Well, that depends on the audience, where they took them, and how they measured this.” Let me tell you practically what we do and how we're getting the results that we're getting. Our kids are learning at 2.3 times the rate. We just got our results from the 2024–25 academic school year, and we're at 2.3 times learning.

The way we measure that is we use something called the NWEA MAP assessment. MAP is a test that's given to about 10 million students in the United States. It's given to public-school, private-school, and homeschooled kids. It's an adaptive test that's able to move along with a child, constantly figuring out their knowledge, as opposed to some state tests, where they just test you on a certain level of curriculum and you get a percentage on that.

There's something called a RIT score, which is basically the score for performance in math, reading, language, and science. Depending on the percentile of a child, their score, and their age, MAP is basically able to say, “Okay, we anticipate that this child will grow along this trajectory.” They might say, “Okay, for a student who is in the 85th percentile in math, we expect this student to go up 4 points over a given year.”

The results we get at our schools show that that student will go up twice the amount that's projected on MAP. Let's say that student is supposed to go up 4 points in a traditional world. We're going to deliver an 8-point increase. What we've found is that we can take a kid, no matter where they show up to us—if they come to us in the 10th percentile or they come to us in the 98th percentile—and deliver a personalized learning experience for that child that meets their needs where they're at and fills in the holes they have. When you see those kinds of results, we can see kids going up multiple grades within a year.

This is another example of just practical knowledge. Take a math curriculum, like fifth-grade math. Generally, in a traditional school, you're spending about 180 days a year in school. You may have a little bit of homework that you're doing, too, so kids are spending probably 200 hours minimum doing something like fifth-grade math. What we've found is that you can complete fifth-grade math in about 20 to 30 hours.

Isn't that insane? This really points to the benefit of one-to-one learning, where kids are going at their own pace. You get rid of so much of the inefficiency of a typical classroom experience. First of all, those 180 hours aren't all spent learning new material. But the other thing is that the quality of the engagement we're able to deliver when a kid is getting that one-to-one mastery tutoring experience is so much higher than sitting in a classroom like this, watching a teacher deliver a certain amount of information over time.

I think one of the responses we get to our results—and we like to be very open with the results we're getting—is that people are incredulous. They're like, “That can't be possible, and you're only doing it in 2 hours a day. How is that possible?” But I think the bigger question is: when we look back at how inefficient the time-based classroom is, it is a joke. That's why only a third of our students in the United States are reading or doing math at grade level.

The problem gets more and more compounded as students get more and more varied in their knowledge. I look at teachers who are doing this most impossible job. How in the world is a teacher supposed to take a group of 20, 25, or 30 kids who are all at such different levels and be expected to get them all through the curriculum they have to do, while also possibly filling holes or challenging the kids who need more advanced information? It's a very sad state of affairs when you look at the way the current education system works.

One of the things that I focus a lot of my energy on is helping people understand that when artificial intelligence comes on the scene, the answer isn't going to be to throw a chatbot on every kid's computer and have the teacher use the chatbot for some extra lesson planning. It really allows us to finally disrupt that teacher-in-front-of-the-classroom model and move teachers into a role where they can actually have a much larger impact, which, for us, we believe is motivational and emotional support and mentorship.

Yeah, there's a lot there. There are a number of huge themes. In terms of just guiding people to understand what's possible with AI, I always come back to a couple of things that you highlighted there. One is just making things scalable that weren't previously scalable.

In business, it's often like, we would always like to be doing more prospecting. We can almost never do enough of that. We would always like to be doing more recruiting, and we can never do enough of that. But how many business leaders are doing as much as they ideally would aspire to? Almost none, right? So, making things scalable that weren't previously scalable.

Also, I usually frame this in terms of cost savings more than time savings, but that 90% savings is another thing that I steer people toward expecting. If you implement a good AI process for whatever process you're trying to get AI to take over for you, 90% savings is usually what I would coach people to expect. So it's interesting that this also translates to how much faster folks can learn. That doesn't come as a huge surprise to me, but it is still a striking result. Yeah.

Well, it's something that I think homeschool parents have known for a long time. It doesn't take 6 hours of sitting in class a day for a student to not just do academics, but actually crush their academics. So that's the other fundamental thing with our model: what we're doing is giving kids back their most valuable resource, which is time to go do other things.

When you walk into one of our schools, we're saying to the kids, "All right, here's what we're going to do. We're going to spend 2 hours in what we call productive struggle—2 hours of focus time focusing on core subjects. And then, starting at noon, you're free to go work on really interesting projects, activities, and group things that they're doing." There's a whole other section where we talk about learning life skills.

But really, what we're doing is providing the motivation for a student to be able to say, "Hey, yeah, I will put my 25 minutes into each subject of focused energy, and then that frees up my afternoons to go work on these interesting projects." There's a whole other component to motivation, which is what our teachers are focused on. And frankly, Nathan, that is the challenge with the majority of, I would say, probably students in our world: if you don't have a motivated student, then there's not going to be a lot of learning going on. Unfortunately, in the traditional system, there's really absolutely no focus put into that aspect of it, which is motivation.

Yeah. Talk about things that are hard to scale. Another one of my quips is, "There's never been a better time to be a motivated learner." And I feel that when I'm motivated to learn something, it's very obvious that AI can teach me extremely well in a lot of cases. I've been using it mostly for biology lately.

But let's come back to the motivation and sort of the afternoon in a minute. I'd love to go quite a bit deeper on the AI-assisted learning experience. Everybody in our audience has used ChatGPT, of course, and Claude, and probably Gemini, especially now that they've got competitive models. Lots of other products, probably, have not used anything.

I personally have used Khan Academy and a couple of other more educationally framed experiences, but what is the experience like? I mean, you said it's not just putting a chatbot on there. That works for me because I'm motivated. What is the experience? How would you describe the experience for the kids?

And, by the way, your comment earlier on the difficulty of measuring—that's another classic theme in AI. My company, Waymark, does creative work for small businesses. Good luck measuring that, right? It's all about taste and what somebody likes versus another person, and there's not a ground truth. There's not even necessarily agreement.

But we can all agree that it's way better to have the AI spit out a decent first draft for you, and in many cases it's better than what our users used to do on their own. So I'm definitely sold that sometimes these things can be hard to measure, and the qualitative experience in the end, in many cases, matters most.

So tell us about the qualitative experience. What's the UI like? What's the experience like? Are there personas? Is this still structured in terms of courses with lessons?

Yeah. No. Let me do that as an explanation, because I think people always want to know, "Okay, you have this AI tutor." We've gotten a lot of news headlines that say, "This is the school with no teachers. They have an AI tutor running the show." But let me give you an idea of what this is like for a student who's going through it.

First of all, when you think about what the components of AI are, the one that most people think of is the chat interface, right? When I go to ChatGPT and ask it for help—I was just filming a script this morning, and I asked ChatGPT, "Look at the script and give me a better ending for this"—that's actually the component that we do not use.

If you go to one of our students at one of our schools and say, "Hey, tell me about your AI tutor," they're not going to say, "Oh, yeah. This little cartoon figure pops up and starts teaching me about math." We don't use that. And here's the reason: if you give a kid a chatbot interface, generally it's going to be used to cheat, right?

We would love the idea that they engage in Socratic discussion with this chat feature, and maybe that's what you're doing with biology. But unfortunately, most of the time kids are like, "Copy and paste: what's the answer?" and then they're putting it in. Or, "Write this essay for me." So that's actually one feature we do not use.

What we're using is Common Core curriculum—K through 8 Common Core, and then, at the high school level, Advanced Placement curriculum. So our students are learning the same curriculum that kids in a traditional school environment are using. We're not having an LLM generate lesson plans. What we're doing is taking the curriculum that's already there.

We're using adaptive apps—things like ALEKS, Grammarly, or IXL—along with a lot of apps that we've built ourselves: Alpha Math, Alpha Read, and Alpha Write. We have another reading tool called TeachTales that we'll talk about. What our kids are doing is using these adaptive apps.

However, when they log into what we call Dash, which is basically our AI tutor and learning platform, they log into Dash and say, "Okay, it's time to go do math." We take that child to exactly the lesson they need to learn within whichever adaptive app the AI tutor has determined is where the child should be working.

When you think about adaptive apps, not all apps are equal at all grade levels or in all subjects. So part of what we've done over the last 11 years is curate which curricula and which apps are best for teaching different subjects, whether that's math, reading, science, language, or reading comprehension.

Unlike in a traditional app world, let's say, if you hand a kid a tablet, they might say, "Hey, I want to go play this lesson. I'm going to do a little bit of this lesson, and then I'm going to back out of it when it starts getting hard. I'm going to go over and jump to this lesson." The experience we have curates what a kid is able to work on. That makes sure that they're always working at the level they're ready for.

Then what our AI tutor is doing in the background is analyzing how accurate and how quickly a child is moving through any specific lesson. It can then provide coaching to a student, saying, "Hey, we notice that your level of accuracy is a little bit low here, and we see that it's because you're not taking time to read the explanation when you get an answer wrong. You're skipping to the next question, so we're going to slow down."

Or it might say, "We're noticing that your accuracy level is a little bit low here. You're taking time to read it, and you're challenged. So we realize that we need to go back and revisit a previous lesson to make sure that you've got the required knowledge you need in order to be successful in this subject."

So what it's really doing is providing a very curated experience to make sure that a child is learning the level of curriculum they need at the pace that works for them, and then putting them in what we call the zone of proximal development. It's this idea that there's a learner's sweet spot where they're challenged to the right level—not overworked, but not unsupported either—and they're able to go with that.

The other thing we're able to look at is something called cognitive load theory. Again, these are a lot of learning-science principles that we can now finally incorporate into a student's learning experience. Cognitive load theory basically talks about the amount of information that we can put in someone's working memory and that they can process at any given time.

Working memory is basically, "Here's what I can put into my head and still manage." If my working memory gets overloaded, pretty soon I will just start short-circuiting. I won't understand. One of the things that's great about this personalized learning experience is that we can say, "Okay, these are the number of concepts that a child can learn about." If they start to get overwhelmed, then they're going to go out of their zone of proximal development, where it's too hard, and we take it down.

On the other end, a kid who can go quickly and learn material quickly needs to get that. I'll give you another example. Let's say it takes you, Nathan, 5 reps of a concept for you to understand it. You shouldn't have to sit through 10 reps if you got it after 5. For me, maybe it takes 15 reps to understand something. I shouldn't only get 10.

That's the other aspect of this personalized AI tutor. What you're going to see a kid tell you if you said, "Okay, tell me about your AI tutor," they're going to say, "Well, I go into my dashboard."

I go to my subject that I'm going to be working on, which is 25 minutes of whatever core subject. It takes me to the adaptive app that's right for what I'm doing. And then I go through the questions, and they're doing everything from video reading to audio support. They're engaged in different things, and they're writing notes on paper along with this.

Basically, what our AI tutor is able to do is continue building out the personalized lessons for where that kid is. It also gives the student, the teacher, and the parent the ability to see literally on a daily basis how much material a student has worked through and how long it's taking them to get through certain lessons. If it's taking longer than it should take, what are the things we need to do as far as coaching them, or where's the challenge that a kid is having?

Really, what we're doing is teaching the kid the skill of being able to learn. How do they gather resources to learn something? Another question that people will often ask is, okay, if a kid gets stuck on something and they're not understanding, what do they do? Let's say they're trying to learn a concept. The AI tutor will give a couple of different examples or explanations: maybe watch this video, read this explanation.

If you're still struggling, go into this resource library and see other explanations that have helped other students who have struggled with this problem. And it'll also say, let's go back and revisit something. So, for example, that kid who maybe is struggling a little bit with fractions, the AI tutor will say, “You know what? We need to go back and revisit the multiplication table.”

And that's another concept called spaced repetition, where we're constantly quizzing kids on the material that they've learned in the past versus, of course, in a traditional classroom, where you learn material and you may never see that material again. You have to hope you remember it. So that's kind of what that experience looks like.

The other aspect of AI that we're using is the vision model. What we're doing is analyzing how a student is working through the material. For example, we provide coaching to a student. The AI tutor will say, “Hey, you're not spending enough time reading an explanation,” and here are examples of that.

We can literally go back and show them the recording of when they were looking at a question, got the question wrong, the explanation popped up, and they moved right to the next question. What is it that you do? What we find there is that's really helping kids take ownership of their learning experience.

Instead of being reactive passengers who show up in a classroom and sit and listen to a teacher talk, and that's it, these kids are learning, “Okay, if I take time to understand this now, it actually helps me go faster and more efficiently through the remaining lessons.”

I guess maybe one high-level question is: You started the school a number of years ago. AI was nowhere near where it is today then. So you've had a chance to ride this technology wave, and you've obviously been an aggressive early adopter of the technology. How has this changed? And you could take that in any number of directions.

But I kind of pinged a friend who's in the education tech and education science space, and he said one huge challenge generally has been that any assessment sort of begs the question of, like, relative to what? We kind of landed on the idea that longitudinal helps a lot, because if you can compare a kid to their past self, that's much better than to some sort of abstract aggregate statistic.

But it strikes me as—and then I was thinking, okay, well, that's interesting. How would I build an AI product to do that? And I was like, well, the easy thing to do now is, with Gemini 2.5 Pro in particular and 1 million tokens of context, I could just gather up a bunch of stuff that the student has done over the last 2 months, perhaps, and say, “Here's everything that this kid has done over the last 2 months. Assess it on these dimensions.”

I honestly think that would work probably pretty well these days. But you didn't have that, right, for a long time. So could you maybe describe the evolution of how you were doing these things, or to the degree you were, pre-AI, and then what early AI unlocked, and then how the latest advances are starting to change how you're doing these things?

Absolutely. A huge difference in how what we're doing has evolved, especially over the last 3 years. But when I started the first school in 2014, I started it when my girls were going into third grade and first grade. The reason I started the school was because I was frustrated with the lack of ability for my kids to get time in a traditional environment to dig into things they were interested in and also really receive that level of knowledge.

Just as an easy example, both my girls went into kindergarten already being able to read chapter books, and yet they would sit in their kindergarten class, and the teacher would be talking about how A sounds like A or B sounds like B. That was just a hard time. Now, again, this is not the teacher's fault. This is a model issue, right? You've got a classroom of kids, and you've got to take basically the lowest common denominator and say, “Okay, we're going to start with assuming you've never seen the alphabet before.”

When I started in 2014, there were apps that were out on the market, right? Khan Academy had already been out there, ALEKS was out there, and there were quite a few different apps that were used. We knew that we could use adaptive apps that would allow kids to jump in at the level they were at.

When my third-grade daughter started at the school, she didn't have to sit and do third-grade content. She could work her way up and do fourth-, fifth-, and sixth-grade content. Here was the problem with those apps: Kids could really bounce around a lot in the apps. You could be doing geometric patterns and shapes, and then as soon as the questions got hard, a kid could hop out of that and go work on something else.

We call them anti-patterns, which are things like topic shopping or not even engaging in that. A kid could certainly just sit on a computer and do 1 question every 15 minutes, right? And that would be inefficient.

The other thing that we found is that when we first started the school, we were letting kids be in charge of where their learning took them. One of the moments when we realized, okay, we need to put some guardrails up, was when we had a first-grade student. He loved math. He was obsessed with math, and he just wanted to do math all the time.

He got to the point where he was doing eighth-grade-level math, and he needed somebody to read the word problems to him because he couldn't read. That's when we realized this kid is not going to spend time reading on his own. We have to make sure we do something.

One of the things that came out pretty early was that we incorporated the Pomodoro technique: 25 minutes of focused attention in each of the core subjects each day. There are a lot of alternative education schools that will say, “Hey, if a kid doesn't want to focus on reading for a few months, no problem. They want to go deep in math or something else, that's fine.”

We took a different approach, which is, again, making sure that students are spending time in each of the core subjects every day and moving along at that level. But there was frankly quite a bit of inefficiency, and kids could mess around.

In 2022, that's when we really saw the power of AI coming out and realized, oh my gosh, we can now use this as a measurement tool. We can use this as an analysis tool to build lesson plans that make sure a student is learning in a linear fashion—the material that they need to learn—and go back and fill holes for material that they don't know.

This is where assessments come in. We can take standardized assessments, which in a traditional environment are kind of pointless and have gotten a bad rap. The reason is because nothing really changes for a child based on how they do on a standardized assessment. Maybe the school gets a grade on how they're doing, but nothing changes for the student.

In our world, we can take an assessment for a student and feed that back into our AI tutor. We build an updated lesson plan that says, “Hey, let's go back and revisit these concepts in each of these different subjects,” and grow from there.

One of the things we saw a big difference in was from the school year of 2021–22 to 2022–23, when we incorporated this AI tutor and personalized learning system and made a couple of other changes that I can get into. That's when we saw our learning rates go from 1.5 to 2-plus. That has been a huge difference.

I can show you graphs of our learning rates over the last several years. There was a major step change when we incorporated AI to make sure we were providing the right level and pace of material, that we were making sure kids weren't wasting time, and that their focus was right.

The other big part of that, which didn't have to do with AI, is that there was some time when we were thinking, well, maybe there's a world where we're still having teachers do some academic teaching. The big change we made in the fall of 2022 was that we said we're going to do no academic teaching.

We are going to allow the AI tutors to do all of the academic teaching. What we found was, when we got out of that hybrid environment, we had learning rates just go like crazy.

And so that's been kind of the big change that happened in the last 3 years. And I think the other key thing that we'll get into in more detail is that for 15 years, edtech has been touted as this great solution that's going to make all the difference, and it's never really worked. There are 2 reasons it hasn't worked. One is, I don't think if you take a kid who's in a time-based classroom, going through 7th-grade math in a school year, and then 3 times a week you have them spend 15 minutes at a time doing Khan Academy or some other app, that really helps. It just plunks the kid back in that same time-based system.

Often, that 15 minutes of time on an app isn't time well used. So I do not advocate for the idea of handing a kid an iPad and a random app and saying, “Let me know when you graduate high school.” Right? It doesn't work. But the other key component to that is the motivation aspect, which is a whole other part of what's made our model work.

So in the 2022 timeframe, at least in language-model terms, that would have been like GPT-3. Were you using language models at that time to process assessments and design?

Yeah. So it started early, and we haven't partnered with any particular LLM. We have really been using Alpha as our testing ground for seeing what the latest and greatest is when it comes to LLMs. I do believe when we get into scaling and how to make this more affordable, there's probably going to be answers where you're able to, again, take LLMs that have gotten less expensive. We burn a lot of money on just the data and analysis we're using, and the vision model of it, right?

Understanding how a kid is working through a problem is giving us the data that we need to understand, 1, how kids generally work through this, and 2, how much time it should take to complete a lesson. Then we're able to troubleshoot if it's taking a kid longer. Why is it? Is it because they literally weren't even engaging with the computer, or because they weren't reading the question, or they weren't watching the video, or whatever that part was?

We haven't partnered with anyone up to this point, and we're constantly testing that out. So it's part of the reason that, again, it's expensive right now. We're burning a lot of AI. I think over time, and I really believe in the next 5 years, we're going to get to the point where every kid can have a tablet for $1,000 a year that's going to allow them to learn all of their information, and I'm excited for where that comes out. And I think it's sooner than we think as far as bringing costs down and what AI is doing for that and allowing us to, again, collect so much information on how students learn that we're going to be able to continue building that into our system.

Yeah, there's no doubt that the cost curve is extremely favorable. Sam Altman just said in the last 24 hours or so that the typical ChatGPT query consumes basically a third of a watt-hour of energy, which, at least in my jurisdiction, with the electricity rates I pay, is like a few hundredths of a cent. So already, the fundamental costs when operating at scale are quite low, and they're forecast to continue to get lower, at least given a fixed level of intelligence. The models that you need for most of this stuff—we have models at that level today, so the cost will definitely—

The other thing that's going to happen—and we're still in the early phases of this—but, for example, we have an app that we created called TeachTales. It's something that anyone can go and use. What it does is it allows us to create reading comprehension at the proper Lexile level for a child—their reading level—and it incorporates their interests.

And so that's one of the other things that's going to be so exciting about generative AI: being able to take the knowledge graph of what a student knows and then overlay it with their interest graph. We do that with some reading comprehension already, but that's going to get even better as we're able to say, “Okay, let's learn math and let's relate it to baseball statistics or fashion design,” and be able to incorporate those things. I think we're just at the beginning of the ability we're going to have as AI gets better at generating curriculum without hallucinations, because thus far we haven't really been focusing on that part. There have been issues with just accuracy, right? If you say, “Hey, give me a 5th-grade math curriculum,” you are going to get some mistakes in there, but that's getting better and it's allowing us to do a lot more with regard to student engagement and having them learn.

We already are able to do that from a reading-comprehension perspective. Kids can give their opinions, like, “I thought that was interesting and I want more articles like this,” and more reading examples that are like this, which is really helpful. That part of engagement and being able to connect a kid—that's another learning-science principle—is, again, the more knowledge that you have in your head, the better you can draw on analogies. That's what allows us to do even more exponential learning experiences, where kids can quickly grab something that they already know and relate it to a new concept that they're learning. And that's where it's only going to get better with AI over the next months and years.

It's one of the things, too, Nathan, that when we have parents who are thinking about putting their kid in our schools, one of the things we say is, “Hey, the only constant is change.” We are constantly rolling out updates to our apps. We're constantly rolling out updates to our learning system. And then we're looking at the data that we have, the measurement data. I'll give you an example of that. In January of 2025, we had our MAP report come out because our students take MAP assessments 3 times a year.

In January, we saw our classes were at the 99th percentile across the board in every grade and every subject except 5th-grade math. We were at the 93rd percentile. So our academic team was able to jump in and be like, “Okay, what's going on in 5th-grade math? We've obviously got some sort of curriculum issue or something going on that's causing us not to perform as well there.” And we can go in and troubleshoot.

That's where I think AI is giving us that precise measurement and assessment. In fact, it's doing it in a more pure environment than if you were in a traditional classroom, where you're like, “Well, I don't know. Was it a bad teacher? Was it bad curriculum? Were the kids not doing anything during the time at school?” What was happening here? We're able to really see all of the different inputs and then more accurately analyze what that output measure is and how we can improve it.

Can you share the big reveal? What did you ultimately find when you got into that 5th-grade math?

That's a good question. I think we realized there were a couple of things. There are a few different apps that are out there. I'll give you an example. There's an app called Math Academy, and Math Academy is a really great math app for kids who are advanced and kids who really like math. It doesn't do as good a job for the average kid who's just getting through math, who doesn't really love math. And so that's an example of something that we found. That's not specific to 5th grade, but in general, we've realized, okay, about 25% of our students really like learning off of Math Academy.

The majority of our students are going to our Alpha Math as a lesson. So sometimes it's everything from, “Okay, this app isn't right.” The other thing we've seen is, when you look at it, you can measure something called depth of knowledge. How far down is an app teaching a student in some subject, and are they getting the right level?

We had an SAT Blitz that we did for our high school students this fall, and one of the apps that we were using, we saw that they were really good on depth of knowledge 1 and 2, but they didn't go enough into 3 and 4. We actually gave feedback to that app company, and that app company rolled out a new version.

So that's the other thing that I think is really exciting about what's happening with regard to innovation and edtech. I think a lot of these app companies really like Alpha because they're getting so much data on how students are using it that, again, hasn't been dirtied by, "Well, we don't know what's going on in the classroom with the teacher and what they've been teaching as far as curriculum." This really is: How are kids learning the material? How are they performing?

That's where I've now got an academic team of psychometricians who are literally figuring out, "Here are all the best ways to help kids learn," both learning science principles as well as the actual curriculum. What are we doing? That's where LLMs and AI in general are going to get so much better at delivering customized content that hits a student where they're at and can start providing things that help them with engagement.

But to be clear, we're already there in certain things. Our high school students are taking their AP courses, and our students created really great, catchy songs and ended up doing a soundtrack that was helping them with their AP U.S. History studying. It was like an updated Hamilton soundtrack that they used AI to feed that in. I remember my daughter, when she was taking AP Art History, singing a song that had the same melody as some Taylor Swift songs to memorize her canons.

That's also just another example of how we're teaching our kids academics by leveraging the power of AI in the morning, but not to cheat. In the afternoons, our kids are learning how to use AI tools to give them superpowers. That's things like our high school students being able to create songs that help them memorize the material they need for their AP tests. And it's a lot of fun.

It's fun to hear, "I kind of know that song. I know the melody, but I do not know those artistic references that are being made." Somewhere, I think at my parents' house, there's still a VHS tape from my AP history course with my rendition of "Stonewall Jackson's Way," which hopefully won't ever appear on the internet.

How do you envision the future of this personalization? It seems like I'm not clear right now on whether you are sticking to Common Core and doing these standard assessments because you need that, because you think that's a good true north, or because you need that for legitimacy. But it strikes me that, in a not-very-distant future, it's at least plausible to think that all this stuff might dissolve and that, at the end of a day, week, month, year, whatever, you might not even need a test.

Or you might not even need a highly structured segmentation between subjects. Is there a more organic future form factor to all this that you can envision, or does that seem too fanciful?

I don't know if I'm going to get the answer to the question that you're asking, but one of the things I think has happened in alt ed is that there's been this movement away from traditional academics and academic knowledge. Often, you'll hear that as, "Hey, it's all about project-based learning." Kids should learn math by running a business and being able to understand that.

I have not had that same philosophy. I actually believe that core academic knowledge and direct instruction is something that we should have in kids' brains, and I believe we have the ability to really download it into our kids' brains so much more efficiently and quickly. I think in order to be a great critical thinker, the more information you have upon which to draw, the better it is.

I definitely don't fall in that camp of, "Hey, we're going to get a little more loosey-goosey on the acquisition of knowledge." I think especially given how easy it is in this new world to get information into a kid's head, why aren't we doing that? First of all, it makes your brain a more interesting place to hang out. That's one of the things I really believe.

Then, getting back to being able to use analogies, when you are building that business and using math for that, you have that basic math knowledge. I think in general, one of the criticisms I've had of traditional education is that standards continue to be lowered. People say, "Hey, let's not worry about things like math facts," but we think it's really important. We even do some amount of drill and kill.

I was walking into our second-grade classroom a few weeks ago, and in the morning we do the morning launch. The kids were doing basically calisthenic exercises, and they were shouting out multiplication-table answers. We want kids to be so fluent in their math, in reading, and in all that stuff. It just allows you to do so much more with your brain when you've got that base knowledge.

I guess my answer to your question is I don't know that I think it's going to be so organic. I do have thoughts that we currently call this 2-hour learning. I think in the near future we could get this to 1-hour learning, where kids are spending 15 minutes per subject instead of 25. In fact, our students generally are spending less than 2 hours a day on learning time.

Our kindergarten, first-, and second-grade students spend more like 90 minutes. Those students are coming out of their experience at Alpha at the end of the school year in the 99th percentile. You can absolutely deliver great knowledge. Again, the beauty of this—and what's going to ultimately make this more scalable and accessible—is that it works for all students, wherever they're at, not just your advanced students or just your remedial students. It can really meet students where they are.

The other reason that I am very transparent about results and measurement of results is because this is something different. Every parent thinks about the way that they were educated, and we were all educated pretty much in a traditional classroom setting. We think, "Well, that's the way it's done." We all had that great teacher who we felt was really instrumental in us having a great learning experience.

I think we need to show that these results are working. We are getting the academic excellence and rigor that we need, and that's what we very much are. We provide rigorous academics, and we believe that every kid has the ability to be at or above the 90th percentile in K–8 Common Core. We also believe that it's a function of putting in the effort and the work, not about IQ or just how smart you are.

There are 2 things needed to learn: You have to have the knowledge given to you at the right pace and level for what you need, and then 90% is that you have to be a motivated student. That's what I think the insight that has made us successful is: putting so much focus on motivating our students to do well in their academics.

What about modalities? Another trend, obviously, in AI has been toward multimodal. The AIs can now see, they can hear, they can talk, and they can generate images. How multimodal is the experience today, and where do you think that is going?

It's one thing to sit through forced corporate training. I've done that, and that's been my most recent edtech experience. You're just clicking through, and I can relate to the kids who are just trying to click through and get to the end of the thing as fast as possible. There are schools of thought that it is actually important to write by hand and that something's lost if you're on a keyboard. I'm not sure if I fully buy that. You might have a stronger point of view.

Is voice involved? Are the kids talking to the AI? Is the AI talking back? Are you actually reading their handwriting off of the pages? What does the multimodal component look like today?

That's one of the things that I think is really cool about this model of learning. There are so many different ways that kids can engage in this learning. Yes, they are able to use voice. One of the places we think about that is in reading, as an example.

It is so important when you're reading and learning to read that you're reading out loud. We use voice-reading techniques where kids are reading back, and then the AI is able to analyze: How are they doing on this reading? How quickly are they reading? Where are they at? Are they at the right level?

That's everything from—we've built an app called Fluency Coach that we use. There's another app called Literably that's available. It's a third-party app that's used. We use these regularly for reading intervention, especially with our pre-K, kindergarten, first-, and second-grade students.

Again, when you think about how important those fundamentals are, in third grade you go from learning to read to reading to learn. If you're not a good reader by third grade, your success from an academic perspective for the rest of your career is not going to be as high. We're making sure we get those fundamentals.

That's one of the things that's so cool about being able to take something like reading and have kids not only practice writing their letters and typing their letters and recognizing them and going on the screen and being able to match things, but then being able to speak out and listen as well for pronunciation. We do phonics-based reading as learning.

So that's one of the things that's cool. We have video and auditory support. It's where we're seeing success with kids who have learning differences, like dyslexia or dysgraphia. Now, to be clear, we are not going in and changing the neural pathways of a child with dyslexia.

There are schools that specialize in helping kids build up tools and rewire their brains to be able to work with that. We generally see that kids who are in the bottom 50th percentile in terms of the severity of that learning difference can do very well in our model. And that is because, again, they're going at the pace that they need. They're getting that auditory and visual support where they need it.

But that's what we do. Handwriting is another example of something that parents, especially when their kids are young, want. And I actually do believe tactile function is important, right? You have to have kids doing that.

If you walk into one of our kindergarten or first-grade classrooms, you'll see the manipulatives that they're able to use as they're doing math. They're getting these little pegs, and they're able to count and lay those out and see that visual representation, feel that, and do that. It's the same thing with being able to practice their handwriting.

In our second-grade classroom, I was in one of the classrooms earlier this fall, and the kids were talking about some of the independent checks that they have to work on in addition to their academics. When someone asked, “What's the hardest check?” they all said in unison, “Cursive is so hard.” So, yeah, we're still doing that.

Now, are they going to need cursive that much in this new world? Who knows? Maybe not. But there are certain things that I think we hold on to and add. Things like writing: our students are doing writing through apps that we have online. We also use independent time for kids to practice real handwriting and do that writing.

And so I think it's important to combine a lot of those different efforts. I think one of the things that's great about these apps and AI tools is that we're able to use a lot of different modalities in order to really meet kids where they're at. Again, sometimes some things don't work for certain kids, and others work really well. That's the other thing: it's truly personalized learning. What works for this kid, and then how can we build our lesson plan around it?

Cool. Well, thank you for spending so much time and going so deep on the technology. Let's talk about the afternoon and the role of the teachers. I think you call them guides in general, right?

We do. We call them guides, coaches, and teachers. We use those words interchangeably. One of the reasons we focused on the name “guide”—and our guides do prefer that as what they're called—is because they are not teaching academic subjects, right? They are not subject-matter experts in terms of saying, “I'm a science teacher” or “I'm the language teacher.”

And so instead of being the sage on the stage, they are the guide on the side. Their job and function are to provide motivational and emotional support and mentorship to these students. We are a very high-standard environment. We believe kids are limitless, and we believe they can be very successful academically. We believe they can do impossible things.

But we also provide the high support that is needed, the mentorship that is needed, to help kids be able to do that. So our teachers are focused on helping kids develop growth mindsets and being able to understand their connection to putting in work and effort and getting results. That is the function.

I believe what we've done is transform the role of the adult in the classroom to focus on what only humans are really great at, which is that personal connection. One of the things that we often say—and we actually ask our students this as they get into older grades—is, “Every adult would say they have 1 or 2 teachers who changed their lives. Do you believe that your guide is that person for you?”

And I'd ask you that question, Nathan. Do you have a teacher or 2 who changed your life?

I can certainly think of a couple.

Yes, a couple.

That's generally the answer. Most people will say, “I had 2.” I had, in my junior and senior years of high school, my business teacher, Mr. Blood, the greatest teacher in the world, and my 6th-grade teacher, Mr. L. They were fantastic.

But then I think about how many teachers I had in my career, and some of them were not good, right? What we're doing is we literally hold our teachers accountable to delivering the 3 commitments that we have for our students: that they'll love school, they'll learn twice as fast in 2 hours, and that they will develop life skills.

For example, when we think about learning 2× in 2 hours, we are handling the part of giving them the correct lesson plans to go at their pace. The other key aspect is motivation. If a kid isn't motivated to learn, we're going to really struggle.

What our teachers do is spend their time getting to know every single kid and having the time to understand, “Okay, this is the thing that motivates this child.” That can be everything from external rewards—for example, we pay our kids, and they earn school currency for getting their work done—to privileges, like being able to sit where they want when they hit their goals. They're able to have more autonomy over their space, get to bring in a Chick-fil-A lunch, earn a trip to the zoo, or whatever it is. We do all kinds of different things to motivate our kids.

An example that I love is a student who would often find himself distracted. He was quite often staring out the window when he maybe should have been working on his different apps. His teacher had enough experience with him and knowledge to know that a lot of times when he was looking out the window, it was because he loves birds. He's super obsessed with ornithology.

So she and the student spent some time coming up with a big poster board of all the different birds that are in our Austin-area greenbelt. When he would hit his goals on a weekly basis, that would earn him 15 minutes of time with his guide to go out into our greenbelt, do bird-watching, and go find all these birds.

That right there was the connection that child needed to say, “Hey, I want to focus on getting my goals done because I get to earn time to go bird-watching,” right? We do a lot of different things in our world, but I think if you were telling me, “Hey, MacKenzie, you're in charge of K–12 public education starting tomorrow,” what would I do for motivation?

I think one of the things that you could do is have kids say, “Okay, you can focus on your academics in the morning, and then you get your afternoons back to go do the things you love to do.” Often, the things those kids like to do are the things they usually start doing at about 3:00, which is football, basketball, theater, robotics class, speech and debate club, dance, or whatever it is.

You can figure out the thing that a kid enjoys spending their time on and give them that time back to go spend there. I think what we generally find is that there are a lot of kids who say, “Gosh, I'll take that focus time in the 2 hours to get time to go do these really fun, interesting things.”

Unfortunately, again, in the traditional model, kids who do well are usually successful based on 2 things: intelligence, IQ, and whether they're willing to be conscientious and do the work. Are they willing to turn in their homework? Are they willing to study for the test?

If a kid's not doing well in a traditional school experience, it's usually considered the kid's fault. It's like, “Well, he's not the smartest kid in the world,” or “He's lazy,” or “He's disengaged,” or “He just doesn't want to do it.”

In our school world, if a kid's not thriving, then it's our fault. In fact, we just got out of our staff days at the end of the school year on Monday and Tuesday, where we went through every single student that we have and said, “Are we delivering these 3 commitments? If we're not, what are we doing to fix it? What is the thing that we have to change in the input in order to get that motivation connected?”

I think it's what turns the model on its head. For our teachers, it's just a much more fulfilling job because they really do have the time and ability to spend getting to know that kid and understanding what makes them tick, as opposed to trying to figure out how they're going to teach curriculum to 20-plus kids.

Speaking of jobs, how do you hire differently? Do you care about education degrees? What advice would you give to people if they want to be a guide at one of your schools? Or even just more broadly, there's this notion that we've gone from physical labor to intellectual labor, and maybe the next big thing could be sort of caring, or the caring economy. Is that the kind of vibe that you're going for? What do people need to know?

When we think about the people who are working in our classrooms—our guides, our teachers—I would say probably about 15% of our teachers come from a traditional teaching background, especially at the younger grade levels. I really believe teachers who are working in classrooms today are absolute heroes.

So people who have said, “Hey, I want to go and spend my time with 5-, 6-, 7-, and 8-year-olds...”

Like, that's who I want in class. We do find that we get a lot of traditional teachers coming into those lower levels, but a much greater majority of the people we've attracted are people who love the idea of education and helping people, but maybe weren't willing to go into a traditional school experience because they're like, "That doesn't work. It's just not where they want to spend their time." They don't see it as being fruitful.

They're really drawn to this idea of being able to provide expertise and motivation, connecting with students, helping students figure out that thing they're interested in, and saying, "Hey, let's go learn about that together. Let's go figure out how to do that."

What we're looking for, first, is really smart adults who are sharp and quick. Kids are so underrated in our society today, and we want adults who really understand, "Wow, kids are so capable. They are really amazing, and given that support and resources, they can do really incredible things." So we're looking for people who buy into that and people who've shown excellence in some sort of motivation.

We get people who come from entrepreneurial backgrounds. We get people who were former professional and college-level athletes and coaches. That's a place where we've found a lot of success with people who come in and know how to create the motivation that sports often does. We get people who've come in from executive jobs, doing different work in the real world, and we find that creates something.

We are hiring like crazy on our guide side, and then we're also hiring people on our tech side. We need people who are continuing to build these apps out and continuing to learn how we can incorporate learning-science principles to develop even better learning results. We do all of that through something called Crossover.com.

The other thing is, we're big fans of paying people well. I really believe that people who choose to dedicate their lives to helping young people in education deserve to be well compensated for that. Our guides start at $100,000 a year and go from there, and I think as a result, we attract people who are excited about coming in and making a big difference.

One thing that's different for us is that we do judge our teachers on how they are delivering these commitments to our students. If they're not motivating students well, then they're not going to be long for this job.

Someone was telling me the other day, they said, "I don't think there's a guide union in your guys' schools. You're never going to have that old battle-axe teacher that nobody likes but is still around." If the teacher is not doing a good job connecting with students, then they're not going to be long for this world. So we're really looking for people who are willing to get in it with kids and really engage.

It's very common that you'll hear our kids talk about how, during breaks and recess, they love that their guides are out playing football with them, playing basketball, and running around on the playground. They are very much people who roll up their sleeves and get in it with the kids. That's what makes them so fun, right?

Again, to be clear, all of the success we're seeing in this model is attributed to our teachers, our guides. I think we've finally equipped them to be extremely impactful by providing the learning platform that allows them to deliver academics in a shorter period of time and much better. But we are by no means a school with no adults in the classroom. Our adults are just able to spend their time in a really impactful way.

Yeah, that's cool. Certainly, running around on the playground is what I like to do most with my kids. Let's talk about economics, ratios, and scalability, and maybe your theory of change. Obviously, you can only start so many schools. I did notice that there's kind of a default price point of $40,000 a year in tuition, but there's also a $10,000-a-year price point in Brownsville, Texas. I don't know if that reflects a cross-subsidy or whatever, but could you give us a little bit of a sense for the economics and how you can see this working? What is the vision?

Maybe you do want to start 1,000,000 schools, but more likely, there's got to be some mechanism for transfer, right? There's got to be some mechanism for transferring the learning that you're doing and the evolution of education that you're pioneering to the public schools that most kids are presumably going to continue to attend.

Absolutely. Most people have heard of our flagship school, which is Alpha School. At Alpha, we are really redefining what parents expect from a kid's private-school experience. We are a high-end private school, so our tuition generally is anywhere between $40,000 and $65,000 a year, depending on the city.

Brownsville is an interesting example. We really wanted to come to that community. Brownsville has one of the poorest school districts in the country, and it has challenging educational options there. They needed a lot of help, and we wanted to be a resource for the community.

It was also an interesting place for us to cut our teeth because about half of the students in our school in Brownsville are children of SpaceX employees, and about half are kids from the local community. We had a lot of diversity, both socioeconomically and racially, as well as students for whom English was a second language.

What we've shown is that those students—in fact, our campus in Brownsville—consistently outperform our other campuses when it comes to their learning rates. I think it's been a great example of being able to show why a personalized-learning platform can deliver such phenomenal results, no matter where a kid comes to us from or what their socioeconomic or demographic background is. It really is the great equalizer when we talk about being able to meet every student where they need to be.

In general, Alpha is spending a lot of its resources and money on the life-skills experiences in the afternoons. Kids do pretty insanely cool things as part of their life-skills workshops.

But we have other schools at lower price points. For example, we have a gifted-and-talented school, a sports academy, and a middle school that focuses on esports and gaming as a motivation model. Those schools are at about $25,000 a year. Then we're rolling out schools at a $15,000-a-year price point.

They're all getting the same academic experience in that 2-hour personalized-learning platform, but what we're doing around the life-skills component is a little bit different.

We've also recently been trying to make a foray into the public world through charter schools. It's been interesting because we've had more mixed reactions. A lot of states have just said no: "We're not interested in letting you take your model of education into the public sphere." But we were fortunate to be approved for a virtual charter school in Arizona, and we're in the process of applying for a physical charter school in Texas.

That will be really exciting because it will allow us to give that kind of education option to as many students as possible. I do believe that, especially as the cost of AI comes down, we can get this to where it is a very financially reasonable option from an academic perspective.

I also believe we're still committed to paying teachers well in this new model. It's really in the afternoons where we're putting things into the life-skills workshops. In our charter schools, kids may not be traveling to Poland to train Ukrainian refugee students on 2-hour learning, but they can be doing sports workshops, entrepreneurial workshops, and vocational workshops as well. That's the big difference in how this works.

I'd also like to serve as an inspiration. I've been able to have conversations with the Department of Education and with the White House around how to make this style of learning more available in the public sphere.

Now, again, Nathan, the thing is, it just takes the public getting really clear and understanding that this is something that adds to education and not something scary. Everybody thinks, "Oh my gosh, when you talk about AI in education, you're thinking about a robot standing in front of a classroom, or you're thinking about kids cheating." Neither of those things is true.

My advice to people in the public sphere has been: don't put a chatbot on a computer and call it a day. Help teachers and parents understand that the role of the teacher is not to be replaced; it's that the role is going to be transformed to focus on the social and emotional motivation side, as opposed to the academic part of it.

I do believe that we're going to get to a point where 1 billion kids around the world are going to be able to take this type of education. As far as what I'm trying to do, I'm trying to build out private schools, starting in the U.S. and eventually internationally, allowing other people to adopt our learning platform in their schools or go start schools where they can have the academics there.

We're doing some intervention services in the public setting for kids who need out-of-class support—remedial or advanced help. Then I think there's going to be an influx of other innovators who come in and build things like what we're doing that are going to help with that.

I spend a lot of my time and energy trying to raise awareness and be a platform for understanding what's possible if you really understand that your kid doesn't need to spend 6 hours sitting in class to do the core academics. They really can crush their academics in a couple of hours a day.

And then the other thing that I think is really spreading like wildfire is how much value parents are seeing in being able to spend a significant portion of the school day working on life skills, financial literacy and entrepreneurship, storytelling and public speaking, leadership and teamwork, socialization, grit, and learning how to deal with failure. Those skills, I think, are what so many parents say they really want, and the traditional education environment doesn't really lend itself to spending time developing those skills. Hopefully, we can get more of that out.

I do a lot on the future of education on social media, showing people what's possible and helping them understand that this isn't scary. Then, of course, the other component that we haven't really gotten into is teaching our students how to use AI tools to be successful. We can have our kindergarten and first-grade students create books, graphic art, movies, and things like that. They're able to code drones and robots and self-driving cars and build video games.

There are so many different things kids can do when they have AI and they're learning how to use it as a tool. It gets us back to the beginning of this podcast, when we said the half-life of knowledge is going to be so much shorter. What kids need to be successful at is knowing how to use AI tools to further their abilities.

It's no longer about the 3 Rs of reading, writing, and arithmetic. Now it's about critical thinking, collaboration, communication, and creativity. If we can give kids the knowledge they need to use these AI tools, that's going to be a huge thing. It's about helping people understand that AI doesn't just mean cheating; it means really growing capabilities.

I'd love to see you come to Detroit, where I live, at some point. But for those who aren't in a position where they can go to one of your schools and enroll today, what advice would you give me if I just want to DIY it to the best of my ability until you or somebody else has such a program?

Well, first of all, we have a homeschool program called Alpha Anywhere. We've spent a lot of time working on that to make sure the motivation model is built into the program, since we don't have kids in a physical building with our guides working with them. We've started seeing the results that we want, and we're going to roll that out on a grander scale.

The other thing is that I think taking time with your kids to try out these tools is really valuable. One great example, for kids at the age yours are, is inspiring a love of reading and helping them get better at reading by using something like our TeachTales app. You can literally go on a website and try it, and suddenly you've got your little boy as the main character in a choose-your-own-adventure story that happens to be about his favorite movie and includes his best friends from his little league team.

It's delivered at the proper reading level. Suddenly, your son is going to be saying, "Let's read more. Let me do more reading." You can use those AI tools to do this. I also think there are a lot of great tools on the internet. For example, I've always been a fan of public speaking. It's something that, with my kids, I wanted them to be great at.

There are all these great AI tools that help kids get better at public speaking in a really unintimidating way. They can get feedback on a talk they give. For your 6-year-old, you can say, "Hey, let's create a story and present this at dinner tonight." Then you can use AI tools to give feedback on their intonation, how to make it more engaging, their pacing, and whether they used filler words.

There are a lot of things to do. One of our students, who's a graduate of Alpha and has just finished her freshman year at Stanford, is named Austin Scholar. She writes a Substack newsletter for parents that gives them information on how to do education in this new world: what some of the great AI apps are and what projects your kids can do during the summer with regard to AI.

Now, one thing I will say, Nathan—and you're a perfect example of this because of the age of your kids—is that this is where I was when my kids were about the same age. I was trying to build all the scaffolding alongside their traditional school experience. I'd say, "Okay, after school we'll do a science club, and we'll have you work on that, and we'll work on writing a book and doing all these things."

Eventually, what happens is that kids get busy, and parents are busy. It's like, "Oh my gosh, how much time do we have to go do an after-school science club, throw in piano lessons, and do all that?" I really do want the resource of time during the day to be something that's more accessible.

I'm launching futureofeducation.net. It'll be live in the month of June, where parents can get free resources for using AI tools and different apps they can try. I think the biggest thing is, as a parent—and I'm sure you're doing this—get familiar with those AI tools.

One of the conversations I've been having with people in the public domain, including people in the administration who are trying to figure out how to become AI-driven in education, is that so many teachers aren't familiar with AI. They don't know how to use it themselves, and therefore they're wary of it. As a parent, start using AI in your daily life to help you with your job and with everything from figuring out what you're going to cook for dinner.

There are so many different ways to use it, and there are a lot of creative outlets that can get kids really excited about building. That's something we really want our students to do: be creators and contributors, not just consumers. AI tools allow us to be great, superb builders of things, and there are plenty of projects that parents and kids can do together to get them excited about that.

Cool. There are multiple pointers there that I will definitely be sure to follow, and I'll look out for the soon-to-be-launched website as well. Outstanding work. I love the vision, and yes, it is time for the future of education.

I'm glad to learn that the future is now at Alpha School with 2-hour learning. MacKenzie Price, thank you for being part of The Cognitive Revolution.

Thank you for having me.