The Peter Attia Drive - #366 ‒ Transforming education with AI and an individualized, mastery-based education model ｜ Joe Liemandt

00:00:00.000 Hey, everyone. Welcome to the Drive podcast. I'm your host, Peter Atiyah. This podcast,

00:00:16.540 my website, and my weekly newsletter all focus on the goal of translating the science of longevity

00:00:21.520 into something accessible for everyone. Our goal is to provide the best content in health and

00:00:26.720 wellness, and we've established a great team of analysts to make this happen. It is extremely

00:00:31.660 important to me to provide all of this content without relying on paid ads. To do this, our work

00:00:36.960 is made entirely possible by our members, and in return, we offer exclusive member-only content

00:00:42.700 and benefits above and beyond what is available for free. If you want to take your knowledge of

00:00:47.940 this space to the next level, it's our goal to ensure members get back much more than the price

00:00:53.200 of the subscription. If you want to learn more about the benefits of our premium membership,

00:00:58.020 head over to peteratiyahmd.com forward slash subscribe. My guest this week is Joe Limar.

00:01:06.480 Joe's a software entrepreneur turned education reformer. As we discussed in the podcast,

00:01:11.360 he dropped out of Stanford in about 1989, started a company called Trilogy that's gone on to become

00:01:17.400 one of the most profitable software companies in the world that you've probably never heard of

00:01:20.760 because it's remained private this entire time. Joe basically left Trilogy three years ago to

00:01:26.720 become the principal of Alpha School, and his passion today, and as he discusses for the next

00:01:32.420 couple of decades of his life, is going to be on transforming K-12 education. Now, of course,

00:01:37.140 this is something that many people have thought about before, but I think what is really remarkable

00:01:41.400 about the way Joe tells the story is that all previous efforts to transform K-12 have been missing a

00:01:48.240 critical piece of infrastructure, and that infrastructure is indeed AI. Now, you may be

00:01:53.080 asking why we're going to have a discussion about education on the drive, but of course, as I state

00:01:59.000 to Joe at the outset of this podcast, you can't really care about science and medicine if you don't

00:02:04.540 care about education. All of us listening today are one day going to be cared for by people who are in

00:02:10.940 K-12 today who are going to have to learn STEM and hopefully be interested enough to choose a career

00:02:17.640 in medicine. So I think we all have a very vested interest in this, not just for the health of our

00:02:24.180 country and our economy, but also on a very deep and personal level. We talk a lot in this episode about

00:02:31.240 what it is that education research has been saying for the past 40 or 50 years that has been

00:02:38.700 unimplementable because of scale. So we talk, of course, in this episode about his path from

00:02:45.000 software engineer to education, why that pivot. We talk about what's wrong in K-12 education and the

00:02:51.220 forces behind it. In fact, just as an anecdote, today, this morning in the newspaper, I was reading

00:02:56.380 something before recording this intro, which I'm recording about a week after the podcast.

00:03:00.760 They just came out with a national assessment of education progress tests, which are tests that are done

00:03:05.620 across the country. We continue to be in a state of decline. So for mathematics, only 55% of 12th

00:03:14.680 graders were able to meet the basic level and only 67% of 12th graders were able to hit that level in

00:03:22.000 reading. Again, these scores represent the lowest scores we've had in over 20 years. So this downhill

00:03:27.680 trend of basic math and reading skills continues. We talk about the case for mastery-based

00:03:34.820 individual learning that's grounded in learning science. Again, I can't say enough about this,

00:03:40.380 and it's something I don't think I really understood until maybe the past year based on my

00:03:45.380 very limited insight and interaction on the fronts of education. But again, this is such a core idea

00:03:51.140 that it's essential to this. We talk about how alpha schools actually run day-to-day, including

00:03:56.260 replacing seat time with roughly two hours of focused academics daily and how kids make up a grade in as

00:04:03.000 little as 20 to 30 hours. We talk about where AI tutors help most and where human coaches are

00:04:09.040 essential. Talk about building core skills and fluency without killing curiosity. Talk about

00:04:14.440 creating motivation systems, goals, time back, and incentives. Talk about how alpha schools track

00:04:22.360 progress and use data to guide learning. Talk about early outcomes, limitations, and what still needs

00:04:27.980 proving. And of course, the roadmap to scale, the costs, the constraints, and how do you bring this

00:04:32.660 to Joe's ultimate goal, which is at least 1 billion kids within 20 years. So without further delay,

00:04:40.740 please enjoy my conversation with Joe Limon.

00:04:48.500 Joe, thank you so much for making time to come over today. It's nice once in a while to have a guest who

00:04:53.160 is local and doesn't have to travel so far. It's great. I super appreciate it. People are probably

00:04:57.740 wondering like, how does this tie into health and medicine? And this is a topic that's a little bit

00:05:02.260 outside of that, but really it's not. Where we're ultimately going to go is education. And you can't

00:05:07.820 have a great system of health and medicine if you're not educating kids today. The people that are

00:05:12.720 learning today are going to be the ones taking care of us. So I think everybody should have a vested

00:05:16.420 interest in this process. Of course, also many people listening have kids across various spectrums of

00:05:22.160 learning. But I think before I get into that, your story is just so interesting. And I know you don't

00:05:27.200 really like telling your story because I know this because I know you so well. And all you want to do

00:05:31.840 is talk about the work, but I'm going to force you to tell your story a little bit. So where'd you

00:05:35.400 grow up? I was born in Minnesota and I moved around every couple of years, mostly up and down the East

00:05:40.080 Coast. My dad worked for General Electric and wherever they had a factory, whatever town we'd move there.

00:05:46.800 You obviously did very well in high school and you wound up at Stanford. So what'd you major in?

00:05:50.940 I would have been, I majored in econ. I did not graduate. I dropped out, but in econ major,

00:05:56.560 I mostly picked it because it was easy in the Stanford curriculum.

00:06:00.760 What year were you in when you dropped out?

00:06:02.320 I would have been class of 90, 1990, and I dropped out between junior and senior year.

00:06:07.560 I'm sure my junior year professors were like, he's not really engaged and probably misses too many

00:06:12.540 classes. For me, and just a little bit on the background was when I was in high school,

00:06:16.660 I wrote a paper on AI. The paper was about expert systems and old school AI, not what we

00:06:21.400 know today. There was a part, one paragraph called Neural Nets, which is sort of today's AI.

00:06:26.100 And it was like, this is decades away. I then went to Stanford and literally I was in a class with

00:06:30.980 Professor Weigenbaum, who's considered one of the fathers of expert systems and old school AI.

00:06:36.160 And he's just talking about how you can build these incredible systems and they'd be worth

00:06:40.280 millions of dollars if you could figure it out. And so some classmates and I, we dropped out,

00:06:45.860 started Trilogy. And the story around that was just, it was the first AI company. We didn't call

00:06:52.740 it AI back then because it had bad rep in the nineties to sell a billion dollars of AI products.

00:06:58.240 And so that's how we got our start and eventually leads to today's story.

00:07:03.340 I always find it amazing when people are dropping out like with one year to go,

00:07:06.180 sort of like, why didn't you finish it? But you must have a strong conviction

00:07:08.880 if you're saying I'm going to forego my Stanford degree to go and start this company that can't

00:07:13.440 wait a year to do this. There's a famous Forbes headline. That's like, you're a moron. Cause

00:07:17.980 my dad was very clear what a moron I was being by dropping out. The issue was I felt there was a 1.00

00:07:23.520 time to market issue. It was a race. And I was just like, I'm sitting in Spanish class. I don't care. 0.70

00:07:29.900 My market's going to run away without me. And there's a huge opportunity to go build this

00:07:33.580 company. It's that tension where you thought now with 2020 hindsight, it took us,

00:07:38.880 over three years to build the product. And we thought we could do much quicker. And so there

00:07:43.280 wasn't a time to market issue. And with, with hindsight, I could have stayed, but back to that

00:07:48.060 pressure, I just felt it. I didn't want to miss it. I had to go do it. Did you guys come and set

00:07:52.780 up shop in Austin right away? Not right away. So we dropped out and did the whole, we couldn't raise

00:07:57.360 funding. Silicon Valley back in the late eighties was not given money to dropouts like us. And, but we

00:08:03.080 literally lived in a garage. You know, we had classmates who had like have a house and we're like,

00:08:06.920 can we rent your garage to live in it? But then in 92, we moved down to Austin. Actually,

00:08:12.820 John Lynch, who was one of the co-founders was like, look, if we're going to be poor,

00:08:17.520 Austin back then Austin was cheap. It's a lot cheaper than Palo Alto. So let's go move down

00:08:21.020 there. And so that's actually what led us down here to Austin.

00:08:25.400 When did this idea of finding your next, is it David Brooks who wrote that book, The Second

00:08:30.800 Mountain? If your first mountain was trilogy, this incredible business, which really we're not going to

00:08:36.360 say much more about other than the fact that it basically created an unlimited, essentially war

00:08:42.260 chest of capital that you were then able to go and deploy to your second mountain, which is this

00:08:46.820 passion around education. Yes, you always had an interest in AI, but when did these two become

00:08:52.460 the next frontier? Back to education 10 years ago, 11 years ago now, McKinsey Price started alpha in

00:09:01.960 Austin. Radical idea back then. One of the things I've learned as principle of alpha is that parents

00:09:07.680 want their kids' education to be however they were educated. And I was the same way. I went to Catholic

00:09:12.740 school. My kids are in Catholic school. And McKinsey's like, look at this great new school I

00:09:17.520 got. It's going to be awesome. And I was like, that's weird. When she started, it's literally in

00:09:23.120 some guy's garage. I'm like, her co-founder, I'm not going to the weird school. Sorry. It took her two

00:09:28.920 years to actually convince me to move my daughters over. They sampled it one time and literally loved 0.91

00:09:35.680 it. They did it after their Catholic school got out before they went to summer camp. And they came

00:09:40.840 back and they're like, dad, we don't want to go to summer camp. We want to still go to alpha. And I

00:09:45.600 didn't like school enough. You weren't going to miss summer for school. That sounded crazy to me.

00:09:50.000 So anyway, I was like, okay, that's actually a really good sign. I should move over. And that was 10

00:09:53.040 years ago. But my discussion with McKinsey always was like, this is awesome, but it's not scalable.

00:09:57.880 What did alpha look like 11 years ago when you first peered behind the curtain?

00:10:03.200 It was the other co-founder, Brian, and he had this philosophy. He had these rules. So when I

00:10:09.060 moved my kid over to alpha, I said, Brian, what are the two things I don't believe

00:10:12.960 about education that I need to understand? He's like, first, kids must love school.

00:10:19.400 They must love school. And I'm like, sort of spinach sometimes, dude. Like, no. He's like,

00:10:24.700 you're wrong and you'll figure it out. And I was like, okay. And then his second is your

00:10:29.780 kids can do so much more than you expect. And I'm like, I sort of have really high expectations.

00:10:36.160 He's like, they're going to blow you away on what they can do. And your expectations are

00:10:41.300 too low. He was setting it up. It really, it was 20 kids. They weren't in the garage after

00:10:46.520 two years, but it is alternative. They would use a set of apps to do academics, but they'd

00:10:53.240 have the teachers would help academically if needed. And then they would have all these

00:10:58.420 life skills. They check charts and life skills. So it was some academics and then workshops

00:11:04.000 and life skills.

00:11:04.800 And the 20 kids that are in there, what grade slash age range?

00:11:09.940 They were probably second grade to fifth grade was probably the range, sixth grade,

00:11:14.460 somewhere around there. And they're clustered. It's sort of like the one room school house.

00:11:18.280 You go in and it is crazy. He's like, oh, they're doing spelling bees. I'm like,

00:11:22.520 they're walking on their hands, doing handstands across the room. And he's like, yeah, watch how

00:11:28.020 they do it. And literally they would have to try to walk across the room in handstands,

00:11:32.800 but recite all the words on their spelling bee is doing it. Could they make it all across?

00:11:38.040 And the kids were just having a ton of fun. And I was always like, are they really learning

00:11:42.400 and sort of all the academic side, but testing showed they were.

00:11:46.400 And what was the state of technology? You referred to some apps, but what was the state

00:11:50.660 of educational apps 11 years ago?

00:11:52.820 There was things like Dreambox and Khan Academy was out.

00:11:56.340 Okay. So they were static, static in the sense that they weren't adaptive.

00:11:59.920 The problem with old ones and actually even a bunch of them today is now we know with learning

00:12:05.380 science, all these much better ways to do it. Their standards of mastery were too low.

00:12:10.620 Their implementation of learning science was too little if you look at them, but they were good

00:12:15.460 enough that you could get your way through. But back to some of the problems, like when my oldest

00:12:21.480 first took her seventh grade test, standardized test, she failed at all.

00:12:26.800 Sorry, this is after being in the alpha program?

00:12:29.820 This was, she had been in the alpha program with those apps.

00:12:32.960 Got it.

00:12:33.240 And so the apps were janky and not all there. This is before I got involved, except as a

00:12:39.080 parent. And she failed her seventh grade standardized test. And you're like, oh, wow, you can use

00:12:44.860 apps. People say, how does alpha get all these results? Because there's a lot of people try

00:12:48.540 to use these apps. And you're like, these apps have gaps and holes. Figuring out how to change

00:12:53.340 that was a lot of how I got involved in what I've been doing the last three years.

00:12:57.760 Your foray into this is obviously then in this logical way through the experience of your own

00:13:02.100 kids. I want to just take a step way back and talk more about education in general. I don't think

00:13:08.140 anybody listening to us now is unfamiliar with the idea, although it's not particularly popular at

00:13:15.980 the moment. We have enough other things that are going on in the world that have people up in arms,

00:13:19.740 but it's sort of cyclic when people look at the data and realize that the United States is not

00:13:25.100 doing well in primary and secondary education. Or I don't know if those are the right terms,

00:13:30.500 but whatever it is up to K to 12. I don't know if secondary is considered college, but we start to

00:13:34.400 do pretty well in college relative to the world. But getting to college, we're not doing well. And

00:13:39.040 it's the point is further made when you look at the dollars in to results out. So I think we're

00:13:44.280 spending what $1 trillion on K through 12 education, which is about one seventh of what

00:13:50.620 the world spends. And of course we don't have one seventh of the population. So there's a huge

00:13:55.060 mismatch there, but our ROI is quite low. So it's actually kind of parallels the healthcare spending

00:14:00.000 thing where on a per capita basis, we spend more on healthcare than anybody else. And yet we're

00:14:04.440 actually not producing the best life expectancy. So let's dwell on that for a moment. Why does the

00:14:10.420 United States spend so much money on K through 12 education? And maybe more importantly, why are

00:14:17.640 we not getting the results? At least if we believe we're measuring the results correctly, if the results

00:14:23.080 are to be determined by these tests that all the OECD nations take, why are we doing so poorly?

00:14:28.260 That's a really complex answer, but it is true. Our academic performance continues to go down

00:14:33.140 across it. As an example, there's this test, NWEA map test. It's a nationwide test. When I was a kid,

00:14:41.440 I took an Iowa test. This is equivalent one. Millions of kids take it. And they just updated

00:14:46.340 their 2025 norm table. The last one they did, it's 2020. You know, every five years they update the

00:14:52.600 benchmark. America knows less now. So your average eighth grader in 2025 knows less than the average

00:14:59.960 eighth grader in 2020, which knows less than the average in 2015.

00:15:04.600 Yep. You can't just blame COVID on that.

00:15:06.360 No, no, no.

00:15:06.660 Like this is a continuing trend.

00:15:08.300 It's a continuing trend. It's continuing to go down. There is one actually anomaly, which is

00:15:12.820 from 15 to 20 and 20 to 25, there's one group who actually goes up and that's your 99th percentile.

00:15:19.760 So the top 1% academically in America in K through 12 continues to increase. Those are your Ivy

00:15:26.160 league bound kids continue to go up. Now, everybody else is going down. Things are getting easier

00:15:32.500 in tests. The AP tests in high school that you take, they re-normed them in the last couple of

00:15:38.000 years and they norm up against college kids because it's like, okay, if you take an AP test, that's like

00:15:42.760 a college class. And so they say, well, what do college kids know? And they're like, oh, college

00:15:47.420 kids know less. So now our AP test is easier. And so it used to be at alpha, if you were going to be

00:15:53.460 in the honors track, you had to get a four or five on an AP. Now we're like, oh, you just need a five

00:15:58.780 because the fours are too easy. The standards just keep going down. And I'll give some more data.

00:16:05.300 I just gave the academic talk. We had hundreds of families at alpha across the country. We have

00:16:10.780 campuses now everywhere start the last couple of weeks. And so they all come in and they take a

00:16:15.700 placement test. I got up in front of all the parrots and I was like, here's the message and it's

00:16:20.940 going to worry you, but you should take away that your $40,000 private school, they've been lying

00:16:26.840 to you, which is if your student has an A and you've been getting straight A's on your transcript,

00:16:32.180 that student in our tests is anywhere from one year ahead, that's great, to three years behind.

00:16:39.320 This is an A student who's three years behind. And when we talk about three years behind,

00:16:43.100 they're a seventh grader. That means they're missing a significant number of questions on the

00:16:47.700 fourth grade test. If you're a B student, you are from three years behind to seven years behind.

00:16:54.900 Let's give an example of what this means. I have a great sense of this because you and I have been

00:16:58.880 through this together, but I want the listener to understand tangibly what that means. Let's put

00:17:03.320 that in terms of something mathematical. Let's talk about what that implies for a B student in seventh

00:17:09.760 grade mathematics. What concept could they be possibly missing?

00:17:15.080 They don't know their multiplication tables or division tables. So one of the things that's

00:17:19.520 happened in K through 12 education is they've decided that memorization is bad. And you don't

00:17:25.700 need to memorize your multiplication tables anymore. You're just going to use a calculator.

00:17:29.240 That's not what learning science says. If you're going to do higher level math, you need fact fluency.

00:17:34.820 And if you think of cognitive load theory, you have a certain number of working memory slots.

00:17:38.940 And the level of complexity of the math problem that you can solve is going to be based on how many

00:17:43.400 slots you have and how you're using them. And one of the few places in brain chemistry is if you

00:17:51.400 have memorized something to fluency, it doesn't use a slot. It doesn't use it. And so you can do

00:17:57.260 higher level math. But if you have to use your slots to say seven times eight is 56, you're now

00:18:03.440 trying to do higher level math and you're using it up on the wrong things. If you want to do higher

00:18:08.320 level math, memorize your math tables and your division tables to fluency. That's the most

00:18:13.480 simple, obvious thing we could do in this country is just change that. Just go back to what we used

00:18:18.140 to do. And even this, and this is the most extreme example. We had a student who had a 740 math SAT.

00:18:24.660 Just for listeners who might not be familiar with that, the score is 800. 740 would be a very good

00:18:29.120 score. It's very good. What percentile is that? Top 10%. Okay. Yeah. 90th percentile. Yeah. 770 is like

00:18:34.240 a top 1%. So maybe in top 5%. Great score. She cannot get above it. We do what we call game

00:18:40.660 film. You review why you miss the test, why you miss the questions on the test. One thing, as an

00:18:45.400 example, that's common in sports, game film. People don't do this in academics for some reason.

00:18:49.620 She would just go through it and she's like, I know what that is. You decide, was it a careless error?

00:18:54.320 Did you not know the material? She's like, after the test, she could get them all done given time.

00:18:58.920 What happened was she was overloaded and made careless mistakes. When your working memory slots

00:19:05.160 are overloaded, it reflects actually in careless mistakes. So if your student's making careless

00:19:09.680 mistakes, you're like grit and discipline. You're like, hmm, you probably don't have a fact fluency

00:19:14.360 somewhere earlier that if you just go fix that, all of a sudden those careless mistakes go away.

00:19:18.640 So she literally went back to third grade, learned her multiplication and division tables.

00:19:22.760 Back to convincing parents, that's really hard. Convincing students and parents, your 740 math

00:19:28.620 student needs third grade material. She then got a 790 math after she memorized it because she stopped

00:19:34.020 making the careless errors. She knew all the material before, she was overloading her working memory,

00:19:39.780 and the learning science is like, that's her issue. In our system, we can identify that and fix it.

00:19:46.700 And seemingly crazy that you can go back and memorize your tables and get up there.

00:19:51.680 I remember another story, Joe, that we bonded over a couple of years ago, which was a student that

00:19:58.400 was really struggling in entry-level chemistry. And I think anybody who's been through chemistry

00:20:04.580 can empathize with that. The first time you're doing, you're going through physical chemistry

00:20:07.940 and you're trying to balance the chemical equations. And this kid is dying and they think

00:20:13.500 the problem is it's chemistry. The problem is chemistry. And you came in and said, no, actually,

00:20:19.060 the problem is fractions. You didn't master fractions in third grade or fourth grade or

00:20:24.340 whatever fractions are introduced. And you're failing chemistry for this exact reason. We're

00:20:30.560 going to go back and fill the gap in fractions and this will become easy for you. And of course,

00:20:35.420 that's what happened.

00:20:36.780 Math science, knowledge is hierarchical, right? And this is back to mastery, which is why we do poorly.

00:20:42.460 We've given up mastery in this country.

00:20:45.520 Why?

00:20:46.240 We're a time-based system. At the end of third grade, you're moving to fourth grade. And at the

00:20:51.600 end of fourth grade, you're moving to fifth grade. And we don't hold kids back. We've said,

00:20:55.360 we're just going to move kids through no matter what. Because it's time. Because there's all these

00:20:59.020 other social issues. Oh my God, I can't have my kid hold back the pressure on the school to relax

00:21:04.660 the standards. 80% of kids at Harvard get an A now. Grade inflation. Grades look like this.

00:21:11.140 Standardized test scores look like that. And for whatever reasons, we've done it. But in a time-based

00:21:16.420 system, just even grading, there's this concept that, okay, you got a 70, you can pass. An 80 is

00:21:22.700 a B. That's great. I mean, kids would be student 80. The problem is if you only have 80% knowledge,

00:21:28.720 you're creating all these holes and it all compounds where eventually, and it starts to

00:21:32.760 when you hit algebra or chemistry, that you think it's the current material, but it's your

00:21:37.900 prereqs, right? It's your foundation. It's your foundation. We talk a lot about parents because

00:21:41.880 a lot of this is, we all grew up in this time-based system and just the old system. And so this new

00:21:46.980 way is very different. But the best analogies we always use that parents totally understand

00:21:51.440 is sports. Just like game film. We're like, review your test, figure out what's wrong.

00:21:55.360 Mastery is the same one. If you're the point guard and you lose the ball 20% of the time going down the

00:22:00.240 court, the coach isn't like, let's work on the advanced stuff. They're like, kid, let's learn how to

00:22:04.960 dribble. We don't need to worry about dunking yet. Correct. No dunking. Let's focus on get up

00:22:10.000 to 99%. Get to mastery that you know how to do the basics. Everybody in sports is like, if you're

00:22:16.060 not doing well, go back and do the basics. Well, academics is the same way. And if you apply that

00:22:21.960 same logic, but in a time-based system with a teacher in front of a classroom, you can't get

00:22:28.120 every kid to mastery. And that's when we get into like the new solutions and why we get these crazy,

00:22:33.000 kids can learn 10 times faster. It's because we're able to get out of a non-mastery time-based

00:22:39.880 system, a non-mastery time-based system that has a teacher in front of a classroom. Teachers are

00:22:45.100 great. Teacher in front of the classroom is bad. That dynamic. And to your point, hierarchical,

00:22:50.520 chemistry is just algebra with advanced word problems. Physics is literally calculus with

00:22:55.900 advanced word problems. Algebra is basically advanced fraction manipulation. You can give kids a

00:23:00.920 fraction test and predict their algebra score. And then fractions are, how's your division and

00:23:05.140 multiplication tables? Do you memorize them to fluency? It just keeps going down the whole track

00:23:09.680 all the way down. And so part of what you have to do to fix the system is figure out age grade.

00:23:17.580 Oh, wait, I'm an eighth grader in algebra. And knowledge grade. Wait, I don't know my fractions

00:23:23.080 are different. A teacher in front of a classroom though, it's not their fault. I'm a teacher. I'm an

00:23:28.540 eighth grade teacher. What do you teach? Eighth grade material. And I have kids who don't know

00:23:32.620 some third, fourth, fifth, sixth, seventh material. What do I do? And this is why so many people hire

00:23:38.260 tutors after school at private schools. It's crazy. We have all these people go to private schools and

00:23:42.760 hire after school tutors because the school doesn't teach them. And the tutor is just like, okay, I'm

00:23:47.760 going to go give you the earlier grade level material. That's it. When we talk about kids, all these

00:23:52.780 hundreds of kids coming in who are three years behind, seven years behind. The second part of

00:23:57.960 what I tell the parents is that's freaking you out because nobody wants their kid this far behind,

00:24:01.880 but it's so easy to fix. The remediation on this stuff is no time at all. When you use these learning

00:24:09.200 science-based AI tutors and the apps that we have, kids can learn 10 times faster, 10 times faster.

00:24:15.520 So when you actually look at material, there's 180 school days, hour a day plus homework, a couple

00:24:22.480 hundred hours. One subject per grade level takes our alpha kids between 20 and 30 hours to finish an

00:24:32.080 entire grade level to mastery. So you can take fourth grade math and be like, you got 26 hours, kid. You do

00:24:39.040 an hour a day of homework. You're 26 days away, right? 22 hours is seventh grade and sixth grade, 22 each

00:24:47.080 for math. The science curriculum, like sixth, seventh and eighth grade. First of all, we don't teach a lot.

00:24:52.700 And Joe, would this work just as a thought experiment if each student had one teacher, the private tutor? If you

00:24:59.520 could put a private tutor in front of a student for 22 hours over the course of whatever, two months or one hour a

00:25:05.860 day for 26 days, are you confident that that individual, and of course, we'll talk about why

00:25:10.160 that's not practical, but that individual, if skilled enough, could navigate and weave that child

00:25:16.320 through? So if you give me a few more hours than what I just gave you, because the AI can do a little

00:25:23.380 bit better than most tutors. But yes, fundamentally, the paper that was written when I was in high school

00:25:28.040 about education, like, so learning science has been around for 40 years. They've been writing papers.

00:25:33.060 There's 10,000 papers published by all the schools of education and Stanford and all over the place.

00:25:38.140 That is, we know kids can learn two, five, or 10 times faster. We've known it. In Bloom's Two Sigma

00:25:44.600 is a seminal paper. And he just said, if everybody got a personalized tutor was his first point. And

00:25:51.260 second, they did it to mastery. They worked to mastery.

00:25:55.220 Right. So mastery-based, not tenure-based.

00:25:57.240 Correct. Just you have to know the material before you advance. And there's an individual tutor

00:26:02.360 to coach you. You will get Two Sigma better performance. You're leftmost part of your

00:26:07.680 curve. Your worst students are going to be at what today is considered the top 10% in a Two Sigma

00:26:12.440 performance. You said to me once, four or five years ago, something, and I'm curious if you still

00:26:17.300 stand by it, which is at least through the end of eighth grade, there is not a single student in the

00:26:24.940 United States that cannot achieve complete mastery in mathematics to score above the 90th or 99th

00:26:33.040 percentile in the national test.

00:26:34.980 Yes. I think I said that to you three years ago, and I would hesitate just for argument's sake. Take out

00:26:40.760 the bottom 5% just to stop all the arguments.

00:26:44.760 Okay. All right. You would still say that 95% of all eighth graders in the United States...

00:26:51.660 ...could be what today is considered top 10% performance.

00:26:54.320 Yep.

00:26:54.540 They can get to mastery.

00:26:55.580 In mathematics.

00:26:56.620 In mathematics.

00:26:57.500 So just let that settle in for a moment. That seems to fly in the face of everything.

00:27:00.980 I want to go back and close the loop on something before we continue this discussion, which is,

00:27:07.200 I still don't have a sense of what's changing, because we've always had a time-based system,

00:27:12.320 not a mastery-based system. So what has gone on in the last 25 years? Because that's not the

00:27:17.120 change. Why are scores getting worse?

00:27:19.060 Standards. Yes. The fundamental answer is standards are lowering. And part of it is parents. Three

00:27:24.160 years ago, when I became principal, I would sit down with parents and I'm like, we have this cool

00:27:28.800 engine. It's going to teach your kid twice as much in the two-hour block, twice as much. Now,

00:27:34.920 they have to use the apps right and engage with it, and your student's not using it right,

00:27:39.940 so they're not learning 2X. The parents are like, stop. Don't pressure my kid to learn 2X.

00:27:45.760 Don't pressure him. I'd be in a parent meeting, and it would literally be mom, dad, and Johnny.

00:27:50.380 And I'm like, look, when he misses a question, he doesn't watch the video, or he's scrolling and

00:27:54.620 just guessing. If he'd just watched the video, he'd learn twice as much. And the parents would say,

00:28:00.060 we're not interested in that. Now, the rest of the things, parents want life skills, community.

00:28:05.620 School is a bundle, of which academics is one piece. And for the most part in America,

00:28:11.300 except for fewer than 10%, less than 10% market share, less than 10% of the parents,

00:28:17.100 fewer than 10% of parents really care about academics. 90% of parents are like, oh, well.

00:28:23.760 So that was an eye-opener to me. Hundreds of parents of meetings later, we switched our pitch,

00:28:29.080 which was two-hour learning. The same engine, you're like, it's two-hour learning. And look,

00:28:34.480 we're trying to get Johnny out of here in two hours so he can do these cool workshops and do what

00:28:38.080 he loves. Parents are like, Johnny, what's wrong with you? Don't you want to get out of here and do

00:28:41.920 the cool stuff? And so the most important part of education is a motivated student. For us,

00:28:47.980 2x learning is not a motivating idea, except for sort of the top kids. Two-hour learning is

00:28:53.880 motivational for every kid on the planet. If you had to say, my biggest unlock was that realization.

00:29:00.140 The product became codenamed Time Back. Give the kids their time back, because that is the most

00:29:06.680 precious for 12 years. We waste kids' time. We know that kids sitting in a lecture-based classroom

00:29:12.340 retain maybe 5%. That bad?

00:29:14.800 It's that bad. It's that bad. Being 10 times better, you're like, I need them to remember 50%.

00:29:19.720 It's that bad. We know it's that bad. It's not personalized to them. It's not to a master's,

00:29:24.680 all these bad things. And so we're wasting their 12 years. And you wonder why kids don't like

00:29:28.860 academics and they're disengaged and all these things is we're wasting their time. And so for this,

00:29:33.200 it's like, great, they're going to learn in two hours. Now they're still learning 2x.

00:29:37.100 That's just not the part we pitch. And so they're learning 2x in two hours. And then you free up

00:29:41.740 the rest of the day. And so we get all our kids. Our score is at Alpha School. We're all top 1%

00:29:47.080 academically. Our classes are top 1% in the country. There's no school academically who's

00:29:52.000 as good as Alpha in the country. And the reason is, is using this model, we're just implementing all

00:29:58.540 these concepts that have been written in these learning science papers for 40 years.

00:30:02.040 Is there a challenge to the intensity that's required in those two hours? So I'll give you

00:30:05.640 an analogy. So I get asked all the time, hey, Peter, look, I've read that I could just do one

00:30:10.420 set to failure once a week for each body part and get most of the benefits of grinding it out for 10

00:30:18.060 to 20 sets per body part. And I say, look, well, it's a little bit of an exaggeration, but there's

00:30:22.420 some truth to this. If you could truly do one to two sets to absolute failure per exercise,

00:30:29.300 yeah, you're going to get great results. And they're like, well, Peter, why don't you do that?

00:30:32.800 And I say, look, to be honest with you, it's really freaking hard to bring that level of intensity to

00:30:38.360 every single set of every single workout every time you do it. And there are a bunch of other

00:30:42.740 reasons that I actually kind of like being in the gym or whatever. But the point is, it's true,

00:30:46.760 you can get great results in much less time. But for many people, the intensity is not something

00:30:52.740 they're willing to rise to. And so I guess the similar question is, if for two hours a day,

00:30:58.340 a kid can learn what a traditional kid is learning in six to seven hours a day and actually do it

00:31:04.380 better, is there something that they have to be able to bring from an attention span perspective

00:31:08.780 that is very difficult? The answer is it's not difficult, but they have to want to do it.

00:31:15.280 So motivation, if you talk to an educator, there's two things you need. You need a motivated student

00:31:20.500 and you need to put them in lessons of the correct difficulty. Not too easy, not too hard.

00:31:25.840 Taking this a little bit, why doesn't EdTech work? Back to these apps and test scores keep going down,

00:31:30.340 but everybody keeps buying EdTech. And the answer is EdTech does that second piece, that 10% piece of

00:31:35.600 put them in the right lessons very well. Our AI tutors do it even better. It can figure out very easily

00:31:40.740 what a kid knows and doesn't know and say, age grade, eighth grade, knowledge grade, fourth,

00:31:45.620 give them fourth grade lessons. So that part's great. Now, the second part is, to the point of

00:31:50.340 my parent meeting, how do I get them to engage with the app? Because I'm spinning in the chair,

00:31:55.820 I'm not paying attention, I'm talking to my friend, I'm scrolling TikTok, I'm doing anything

00:31:59.500 except focusing on it. And so what is their motivation is 90% of the solution. And for us,

00:32:08.060 that's why time back is so valuable, which is the kid has to engage with the apps for those two hours

00:32:13.560 or they don't get the rest of the day. They don't get their time back. They'll just sit there wasting

00:32:18.800 their time. I mean, we literally in our app have a waste meter and it just coaches the kid. We can get

00:32:25.040 into the AIs and stuff and chatbots are terrible. ChatGPT rolled out to America students, 90% of kids

00:32:31.120 use it as a cheat bot. It's all cheating. We don't have a chat interface in ours. But one of the things

00:32:36.620 our AI does is it watches the kid's screen and coaches them up and is like, look, you're scrolling

00:32:42.720 the article and guessing. I'm not going to learn that way. You're skipping this and it's going to

00:32:47.880 coach them of how to learn better. And if you have a motivation for the student to actually learn

00:32:52.860 better and be effective, they're very happy to do that. We have a sports academy that we just opened

00:32:58.540 10 of them around Texas. And these are middle schools, D1 bound athletes kind of, they come

00:33:04.800 in and they do the two hours in the apps. And then once they hit it, they go green where they've done

00:33:10.460 the lessons. They get to do sports all afternoon. And so you can take kids who are like, I hate school.

00:33:15.440 I skip school. Absenteeism is huge. And when they get to do that, all of a sudden they're like,

00:33:20.140 I'm totally motivated. It's only been a couple of weeks that we've been open, but you get notes

00:33:23.960 from all the parents who are like, my kid always was sick, never wanted to go, everything. Now he's

00:33:28.100 waking me up and is like, let's not be late. It's because they have the motivation to actually

00:33:33.040 learn. So that's the most important. You can do that for everybody. And motivation really,

00:33:37.400 really does matter. And alpha is designed around how do we motivate kids. And the biggest motivators

00:33:42.760 time back, we also use a lot of other ones. That's the biggest thing that makes our system work.

00:33:46.980 Then when you talk just, is there also extra juice of the intensity, right? We measure,

00:33:53.240 do kids get through our apps 90 or 95% first time, right? Where we're not making the lesson

00:33:59.960 hard enough, where we want to make sure the lesson's not too hard, where you don't have

00:34:04.120 to have too much intensity. What is the sweet spot? Is it 90?

00:34:08.140 Between 90 and 95% is where first time, right?

00:34:11.480 Got it. Okay.

00:34:11.980 So actually there's a thing called the zone of proximal development, which is what an individual

00:34:16.760 student should be. And you want to keep a student in 80 to 85% correct questions, which is different

00:34:24.040 than first time, right? I'm getting through the whole lesson. You want to be in the zone of

00:34:27.000 proximal development, which is if you're at 99%, it's too easy. You're not learning. If you're

00:34:32.140 getting 99% of the questions, right? You're just regurgitating what you already know. If you're

00:34:36.240 missing half the time or two thirds, you're only at 50 or 60, you get disengagement. It's

00:34:42.280 too frustrating. I'm out. Video game developers are the world's experts on this, on engagement

00:34:46.660 of how do you engage? And they'll tell you always winning, always learning. That concept

00:34:53.100 in 80 to 85% is where that is. So we're measuring the way to maximize learning is a unending stream

00:35:02.260 of content, both educational questions, right? And learning content that is at 80 to 85% accuracy.

00:35:09.180 And this gets into stuff that we're building is if you sit in a Stanford lecture, you're going to

00:35:13.780 retain one to five facts, ideas, and concepts per hour. Our current learning engines, depending on

00:35:19.900 what the subject, we're at 20 to 40. We have stuff that we're working on. And Gen AI is the driver of

00:35:26.020 this. We'll get to a hundred facts, ideas, and concepts an hour. Think of an endlessly scrolled

00:35:30.880 TikTok list, except engaging and good content instead of the poison that that is. And so the

00:35:36.440 learning science part is we know how to get knowledge into kids' head 10 times better than

00:35:43.420 the current educational system. It takes a whole rebuild and back to kids when they love school,

00:35:49.420 they're going to engage in the app because the whole day is great.

00:35:52.260 Let's go back and talk a little bit about education science. You've alluded to it a couple of times.

00:35:56.220 Tell me what 90% of people would agree with. I was going to say, tell me something that is

00:36:01.480 unambiguously true about education science, but let's soften that a little bit and say,

00:36:05.180 tell me something that 90% of people who study education, study the science of learning would

00:36:11.720 agree with.

00:36:12.960 90% would agree that individualized tutoring and a mastery-based standard is the best way to teach

00:36:21.720 kids. It's dramatically more effective than a time-based system.

00:36:24.500 The concepts in learning science, zone of proximal development, spaced repetition,

00:36:30.680 you would get the testing effect that actually asking kids, active learning is better than

00:36:36.300 passive learning. Retention of sitting passively in a classroom is the worst way to teach a kid.

00:36:42.020 All of those things. What else are really well-normed? Cognitive load theory is a newer area of

00:36:47.480 learning science that started in the 90s. And this is the, how many working memory slots do you have?

00:36:52.720 And then how many reps does it take to move from working memory slot, short-term memory,

00:36:58.020 basically working memory slot to long-term? They're all going to agree.

00:37:01.200 It's so interesting to me that there's any dispute around that when you see the effects

00:37:04.420 in sports. Sports illustrates that I think more than anything else. You look at the difference

00:37:09.420 between a Tom Brady versus an exceptional quarterback, a very good quarterback. And it's

00:37:16.820 the processing speed. It's not how fast or how far he can throw the ball. It's his processing speed

00:37:21.500 that sets him apart. It's the cognitive bandwidth of what he's doing. I mean, it's actually the same

00:37:27.340 in formula one. If you look at what separates the very best in the world from very good,

00:37:34.460 it's cognitive bandwidth. It's the guy who actually doesn't have to think at all about driving

00:37:39.520 and can still drive at the limit while he can use all of his cognitive reserve for strategic

00:37:46.260 decision-making. So like that fluency stuff, there's nobody in learning science who's like,

00:37:50.760 it's better if they don't memorize their multiplication tables, even though our school

00:37:55.180 system, it's not like, oh, these private schools are bailing on it too. It's not just...

00:37:59.580 But why was that decision made? Parental pressure again?

00:38:02.360 My kid is not happy.

00:38:03.860 So this is another one of the sort of coddling of the...

00:38:06.440 It is. To come to an alpha school now, I have a conversation with parents. I've sort of

00:38:10.160 fried what he gave you, those two things. But I've had a third one to it, which is,

00:38:14.340 first, your kid must love school, really love school. And we measure, literally measure,

00:38:19.060 would you rather go to school or go on vacation? Super high standard. We should have kids love

00:38:24.180 school for 12 years. They're going to be there your childhood. Why do we not have that standard?

00:38:29.020 The second is your kid can learn in two hours twice as much as sitting in class for six.

00:38:34.420 Your kid can crush academics in just two hours a day. Crush it. That's the second thing you have

00:38:39.840 to believe to come to alpha. And the third, and this is where the controversy parts,

00:38:43.740 but it's true, is the key to your kid's happiness is high standards. And that's the

00:38:50.320 one that a lot of people do not believe, that they think the key to my kid's happiness is low

00:38:55.400 standards. Where did this virus come from? Where did this virus infect us?

00:38:59.900 Well, part of it is the parents want to protect their kids. It comes from parents from a good

00:39:04.960 place, which is, I don't want to see my kids struggle and fail on their road to success. They

00:39:12.200 don't want to see that. I talk about the cycle. As a principle, I can tell you kids struggle, fail,

00:39:17.840 sometimes cry on their road to success. If they're supported by a caring, loving adult,

00:39:23.860 that loop, if you ask any child development expert in the planet, you can type that all in chat to BT

00:39:28.540 is the key to kids development, the key to their self-confidence, the key to their resilience,

00:39:33.200 the key to their growth. They go through a cycle of struggle, fail on the road.

00:39:38.360 It's a hero's journey. It's woven into everything we've done.

00:39:42.440 And back to your point on sports, 50% of America is like, you know where my kid learns life skills?

00:39:46.500 And when you talk about life skills, teamwork, leadership, grit, hard work,

00:39:50.100 it's an afterschool sports because that's what the coach does. My kid's going to go through that

00:39:55.620 struggle. And I went through it when I was playing sports and my kid's going to go through that.

00:39:59.600 And that's where you learn it. And they have this view of that's the only place you learn it.

00:40:04.360 And it shouldn't happen in school. And the reason it doesn't happen in schools

00:40:08.520 is because in a time-based system, everybody thinks good grades equals high IQ. And so it's

00:40:15.140 fundamentally that IQ is what determines grades. And so I don't want to pressure my kid because

00:40:22.500 it's not fair. And so I don't want to put that pressure on. One of the magic unlocks we have,

00:40:27.820 back to everybody can get to much higher performance, is a mastery-based system.

00:40:34.100 If you take the best kids, you're like, who are the kids who do get in the Ivy League?

00:40:38.000 They're basically the high IQ kids who also have high conscientiousness, big five conscientiousness.

00:40:43.360 They're grinders. They just crank through it naturally. And those are the kids who do well.

00:40:48.500 And so if we're rebuilding education to fix it, we have to take those two constraints away.

00:40:53.400 A mastery-based system takes away IQ, which is every kid. The conceptual leaps that are required

00:41:01.840 to get through K through eight, actually by the design of Common Core, aren't that big. That does

00:41:07.400 not mean everybody can go be a nuclear physicist at Stanford and design all these things. There is

00:41:13.720 areas where you do need more cognitive. Everybody in math, there's a point they reach. Just,

00:41:19.720 I can't do the processing. But K through eight, it's designed that everybody can do it.

00:41:25.360 Everyone will do it.

00:41:25.960 Everybody can do it.

00:41:27.140 Just reflect on that for a moment, Joe. Think of all the kids that are in sixth,

00:41:30.880 seventh, and eighth grade who are good students in other domains. They're doing well in English.

00:41:36.380 They're doing well in history. They're conscientious. They work hard,

00:41:39.840 but they've created a narrative that says, I can't do math.

00:41:43.820 We'll dive into that. We also use our motivation model to undo the conscientiousness limitation.

00:41:48.680 You have to use more extrinsic motivators to let people who aren't high conscientiousness

00:41:55.200 still perform like somebody who does. So those are our magic unlocks on those two things that we

00:42:00.480 think can fix education. But your math thing, I got to give these examples. So it's an epidemic of

00:42:05.640 kids who come into middle school, predominantly girls. I have two daughters who come in and they're

00:42:11.380 like, I'm not a math science girl. Society's like accepting it. Okay, let's go read all the learning

00:42:15.780 science papers about how middle school girls can't do math science. Okay, there are none.

00:42:19.460 There's none. That doesn't exist. There's no such thing. But it's this total meme that they believe.

00:42:25.160 And so part of our job in middle or when they transfer in high school is to fix that.

00:42:29.220 There's this concept of, oh, I'm an art girl or I'm an English girl or whatever it is. And it

00:42:33.280 doesn't, girl, boy, there's no difference. We have some seniors this year. The first one,

00:42:38.640 her big project that she does at Alpha, you do these big Alpha X projects. She is doing the first

00:42:45.260 all teen produced, created musical for Broadway. She sourced all the songwriters off TikTok. She can

00:42:54.080 fill a thousand person venue. She's trying to raise it up so she can do it. So she'd be, you know,

00:42:58.500 I'm an arts theater girl. 790 math SAT. We have another girl. These are seniors. She's trying out to

00:43:05.860 be Miss Teen USA. 790 math SAT. Third girl, this is my daughter. She has a, I call her TikTok girl.

00:43:14.980 She's got millions of likes on TikTok and tons of followers. She's building a app to teach teen boys

00:43:20.380 how to ask girls out on dates because teen boys are terrible dad. Top 1% SAT. There's this concept

00:43:26.040 that these people can't learn it. You're like, they all can. Now, what do you have to do when they come

00:43:31.500 into middle school? They do. They don't know the multiplication tables. They don't know the

00:43:35.380 prereqs. They are going to struggle in algebra. So you're like, okay, let's go back. Let's fill

00:43:40.420 them up. And all of a sudden, once you fill the holes, all of a sudden they can do it. And then

00:43:45.480 once you know algebra, you're like, oh, I can do chemistry. K through eight science, there's no

00:43:50.500 math. Actually, it's all just memorizing facts, igneous rocks and planets and stuff. Once you get

00:43:56.120 to high school, that's where biology is still more memorization, less math. But then chemistry and

00:44:00.580 physics are all math-based, calculus and algebra. You have to get these kids the prereqs. But then

00:44:06.080 all of a sudden, they're like, oh, chemistry is pretty fun. Now that I know calculus and I can

00:44:09.900 crush Calc BC, all our kids in high school, if you want to be on the honest track, it's five on Calc BC.

00:44:15.140 Totally doable for everybody. This whole thing, you know, our freshmen coming in,

00:44:20.060 by the end of the freshman year, they're 1410 on the SAT, which 1350 is top 10%.

00:44:25.180 These are freshmen. These are freshmen. And most schools don't, I mean, that's better than most

00:44:29.020 junior seniors. Most don't publish their freshman results. And we can get freshmen there. Once it

00:44:35.040 becomes this mastery-based system, how high you go becomes a decision back to your effort. It's a

00:44:40.780 decision, not an inherent capability. And so when you break that link, alpha kids are just in a mastery-based

00:44:48.460 system. They're like, oh, it's just more work. It's not, I'm not smart enough. And that is the

00:44:54.400 unlock that we have to have and why we have to move to a mastery-based system. Because you just

00:44:58.480 need kids to be like, oh, if you're in normal alpha high, 1350 is the minimum SAT. You got to get to

00:45:04.300 it. Top 10%. If you want to be in the honest track, it's 1550.

00:45:08.200 1550. I mean, that's top 1%.

00:45:09.840 Yeah. Oh yeah. It's just a decision. Do you want to do the work to get to 1550 from 1350?

00:45:15.320 It's just that, look, I'm willing to put in this many more hours this year.

00:45:19.180 Yeah. Back to my youngest daughter, just took the SAT. She's top 1%, but she's not perfect 800. 0.99

00:45:26.620 She's asking the AI, she's like, how much time will it take for me to get up to the top to a

00:45:32.920 perfect 800? And at that level, it doesn't really matter for getting into college, but she's like-

00:45:39.260 Just out of curiosity.

00:45:39.980 Well, no. She's like, dad, all those STEM boys, they think I'm this dumb TikTok girl. And you 0.98

00:45:46.140 know, if I get an 800, I get to mog on them the rest of their life. And I think it's totally worth

00:45:50.700 it. And so she's literally, but it's just a decision. She's trying to decide, it'll come back

00:45:54.680 and it'll probably be 20-ish hours probably, I don't know, some manageable number of hours. And

00:45:59.780 she'll just be making a decision. Am I going to put the reps in, the practice in to get there

00:46:04.260 so that I can have this ego boost, basically, because she really loves it? Or no, okay, I'm

00:46:10.400 going to do something else with my time. That's where in these mastery-based models versus teacher

00:46:15.520 in front of a classroom model, where you're a time-based system, you're like, I don't know how

00:46:18.940 to get there. There's no path up. And when you log in to our system, it literally tells you,

00:46:25.800 you know, in two hours a day, you're three years behind. So you first come in, you're like,

00:46:28.960 you're three years behind. You're like, oh my God, that seems insurmountable.

00:46:32.620 You're 20 to 30 hours per grade level. So you could be 60 hours behind if you're three

00:46:37.780 years behind. 60 hours, that's it. You're 60 hours away. And you're like, so if I do an

00:46:43.580 hour of homework, I'm caught up. Now, back to incentives, there's some kids who are like,

00:46:49.720 do I care that I'm caught up? But it's easy to incentive to want to catch up. You can figure

00:46:55.660 out, okay, what do we have to do? One of the things Alpha is known for, whether infamously

00:47:00.680 or not, is we use extrinsic motivators. Use money. Time back is the most important one.

00:47:05.840 It's more important than money. It's the Westlake track, honors track, where it's six hours a day

00:47:10.440 and eight zillion homework hours. I don't think you can pay a kid enough to love school by going

00:47:15.860 through that. Versus we're two hours a day for a non-honors track. We're three hours a day for

00:47:20.620 honors, back to the 1550. It does take more time. But we will take middle school kids. This is a

00:47:26.700 middle school active program we have. Kids come in. We will pay them $1,000 to get to top 1%.

00:47:33.680 There's a lot of stories around this where kids view, look, I can't do that. I'm not the smart

00:47:40.780 kid. I'm not the math science girl. I can't, whatever it is. And if you get there and do these

00:47:44.900 lessons and we're just going to keep giving you lessons and you do the work, you're going to get

00:47:48.160 $1,000. We'll pay kids coming in middle school $1,000 to get to top 1%. The point of this is what's even

00:47:56.400 more important than the academic knowledge. It's their internal view of themselves that changes.

00:48:02.820 That kids come in and say, I'm not a math science girl. I can't be top 1%. I can't do this. Whatever

00:48:07.540 limits they have. And a lot of education when you talk about it is holistically, you're trying to

00:48:13.260 take limits in kids and release them, have them understand you're limitless. You can do all this.

00:48:19.980 And this is the best way academically we've found to do it. And so they come in and the AI is going to

00:48:25.500 say, here's all the lessons you need to do to get to top 1% and just go through them. And that

00:48:30.960 motivation will get all these kids to top 1%. And the number of stories we have, you know, I have

00:48:36.820 one of my own daughter, but all the alpha families where their student, they come back and they're

00:48:41.600 like, I didn't realize I could be top 1% and my personal self-confidence and view of myself that I

00:48:47.840 had to do hard work. You had to do the work. And then once you take a middle school kid who's top 1%,

00:48:52.640 as they go into high school, they're not going to give it up. And there's this mental view of,

00:48:57.520 oh, I guess I am a smart kid or I'm a capable kid. And that unlock is so valuable. So for us,

00:49:04.920 that's when we think juicing it with money. I mean, we spend 20 grand a kid in this country

00:49:09.520 per student. And you're like, in middle school, give a kid $1,000 in order to have them realize

00:49:15.500 there's no limit.

00:49:16.420 Why would most people, I assume, listening to us have sort of an uneasy feeling about that?

00:49:22.680 They can't reconcile what you just said, which is, well, wait a minute. On the one hand,

00:49:26.400 I know we're spending $20,000 per year per kid on these kids and we're getting nothing for it.

00:49:31.040 And you're proposing, we give them a couple of those thousands of dollars strategically to

00:49:35.620 incentivize half of the equation, which is them having the motivation to do this. And yet,

00:49:41.020 that's going to make a lot of people feel uneasy. Why? So there's a view that there needs to be an

00:49:47.160 intrinsic love of learning for it to be pure. It has to be intrinsic, not extrinsic. And that if

00:49:54.700 you're buying it, it's sullied and not good. It's wrong. It's fundamentally wrong. That is a common view.

00:50:02.700 Are there any data that would support one view being favorable to the other? How would that be studied?

00:50:07.460 There is a set of papers that have been written that show that if somebody is not intrinsically

00:50:13.320 motivated or is intrinsically motivated and you pay them, they lose intrinsic motivation.

00:50:17.640 Now, that's not the case that we're sitting with. We have kids who aren't motivated, who aren't

00:50:23.300 intrinsically going through it. They need the boost. And if you actually ask alpha kids, and maybe we'll

00:50:29.980 do a study someday, no parent who has their kids at alpha thinks, my kids lost their intrinsic

00:50:37.440 motivation. If anything, it's the opposite because now they think they're capable. And those papers

00:50:42.240 don't talk about is how much of the issue was, I didn't think I was capable. I now realize I am.

00:50:49.100 And I get this crazy intrinsic boost around it. My youngest daughter would be this example,

00:50:56.300 which is she came in to middle school. She hit top 10%. This is going into seventh grade. She was top

00:51:03.200 10% coming in. So she'd been there, but she's like, dad, I'm not like my older sister. I'm not top

00:51:08.560 1%. That's just not who I am. I'm TikTok girl. I was like, you know what I know you know love more

00:51:15.080 than anything? It's shopping. I'll bet you $1,000 that you can't get to top 1%. And so she literally 1.00

00:51:22.620 would fill her Amazon cart with $1,000 of stuff. And in the morning, she'd then look and look at the

00:51:28.040 lessons she had to go through and she'd do it. She'd just grind through it. Now, back to this

00:51:33.880 thing. We had a lot of family discussions around this because my wife, she was like, wait, there's

00:51:39.240 two things. Most parents are like this. There's two things here. One, you're setting 1%, top 1%.

00:51:45.420 And most parents are like, that's too hard. Why are you putting that pressure on the kid? And number

00:51:49.820 two, it doesn't feel right to pay a thousand bucks to get this. We were all brought up that there's

00:51:54.200 something wrong, but she does it. She hits top 1%, buys everything. But then she sits down with my 1.00

00:52:00.680 wife and she's like, mom, I know you didn't like us doing this, but you need to understand what this

00:52:06.920 has done for me, which is I never thought I was as smart as my older sister. And now I realize I can

00:52:13.460 do everything that she can do. And it's totally changed my internal view. And that element now,

00:52:21.120 in fact, she's also the one now who's top 1% on the SAT, tried to decide if she wants to do the 0.72

00:52:24.720 extra work to get a perfect score. That change in their view of their capability, that unlock is so

00:52:30.740 valuable. And I believe we should do this for every kid in America. But that's the dynamics of why people

00:52:36.020 don't like it. You then just have to decide if you can use extrinsic motivation, whatever it is,

00:52:43.020 buy them concert tickets, go shopping, screen time, video game time, whatever it takes

00:52:47.440 to get the kid to realize I can go be and learn all this and be top 1% when I previously didn't

00:52:53.740 think that. And I go through that struggle, fail, sometimes cry out of the road to success loop

00:52:58.780 that then unlocks my human potential. The point of education beyond academic, you're trying to

00:53:04.520 unlock human potential. And this is just a great, easy way to do it. It's not available in a time-based

00:53:10.160 system. That's the thing. It's not necessarily something that has to happen indefinitely.

00:53:13.980 What I think makes sense here is that it's an unlock that then creates a new state of awareness

00:53:21.060 that therefore becomes self-perpetuating. So in many ways, it's like the kindling

00:53:25.540 that leads to the fire. You don't need kindling indefinitely, but sometimes you need kindling.

00:53:31.520 I'm going to steal that from you because that's a great way to think about it, which is I'm not

00:53:35.320 paying her to decide to go get a perfect score on her SAT. It's intrinsic. It's her decision. It's

00:53:40.280 her capability. She just needed that kindling to jumpstart it. Whatever block was in the mind,

00:53:45.560 good coaches, they're like, how do I unblock them? And that's what this does. Since we're on this of

00:53:49.600 paying, I'll give you my second one. The secretary of education is coming down next week to meet with

00:53:53.900 McKinsey around alpha and stuff. And if there's one thing to fix education in America, what's the

00:53:59.500 simplest unlock? It would be this. And it's all around whole filling and paying kids, which is,

00:54:05.720 so when I first started, I wanted kids to go back and whole fill. Call it whole filling. Go back and

00:54:11.080 do your prereqs. I'm sitting down with seventh graders and you know what does not motivate a

00:54:15.960 seventh grader? Doing fourth grade. No kid wants to do that. No one wants to be going back. And the

00:54:21.140 only people who want it less than the kid are their parents. Every parent in America has been trained,

00:54:26.220 my kid needs to be at grade level, which once again, the learning science, that is not right.

00:54:30.560 Get your kid to master the basics. And if it's previous material, give it to them.

00:54:34.640 And I tried to convince parents, your kid needs a whole filth. We're going to get this thing,

00:54:38.100 whole thing working and stuff. And I basically ended up giving up on the parents. And so then

00:54:42.040 I went to the seventh graders and I was just like, okay, here we go, guys. You could get a hundred on

00:54:46.420 the Texas star, you know, our state test. They're like, no way, Mr. Lee Watt. I'm like, no, no,

00:54:52.160 no. In century, I'll give you a hundred dollars bill for a hundred on the test. They're like,

00:54:56.720 that's still impossible. I'm like, no, no, no. There's a catch. Any grade level. They're like,

00:55:02.320 I can take a kindergarten Texas star and you can give me a hundred bucks. I was like, well,

00:55:06.340 it starts in the third grade. They're like, I'm in. They take the third grade Texas star,

00:55:11.700 they get a hundred. I give them a hundred dollar bill. They're like, a fourth grade. I'm like,

00:55:15.700 absolutely. Same thing. A hundred, get a hundred.

00:55:18.360 By fifth grade, they're running into trouble.

00:55:20.100 75 to 85. Right. They get 75 to 85. There's holes. I'm like, well, do you want the AI tutor

00:55:26.340 to generate the lessons you missed? And they're like, well, let me see how many of it is. They're

00:55:30.780 like, I can do that in a week. They're in. Get the hundred. On the way, sixth and the seventh grade.

00:55:35.740 And for $400, $400. You brought a kid up to grade level.

00:55:41.580 Up to grade level. But more importantly, they all think and know they can get a hundred on the state

00:55:46.380 test, which used to be only the GT kids and smart kids get it. If you survey incoming alpha parents

00:55:52.000 and say, can your kid get a hundred on the Texas star? Fewer than 10% say, yes, they can. After

00:55:57.440 you've been at alpha a year, 90% of kids are like, I can totally get a hundred on Texas star because

00:56:02.780 they've done it right. Our building over there gets more hundreds on the Texas star than a school

00:56:09.900 district of a hundred thousand kids. And it's just because we have a mastery based system where

00:56:14.400 our advancement now and is built into our system is we'll actually advance a kid. If they get over

00:56:19.600 90, what is our mastery standard? It's in the nineties. So you have to score on the state test

00:56:24.580 above the 90 to move on to the next grade level material. But we also have a hundred for a hundred.

00:56:30.180 And so the kids are like, okay, well, I guess I'm just going to get the hundred because I want to

00:56:33.620 get the hundred dollar bill. So you have this concept where, and it just changes them where I could

00:56:38.120 advance. That's an interesting experiment in and of itself. It's sort of what we would call a natural

00:56:42.320 experiment because it's not randomized, which is if the kid is motivated, basically just to do the

00:56:48.340 bare minimum, they're going to just take the 90. Right. And if you make it a hundred, there's a

00:56:53.460 little reality of bringing your parents and everybody along with you on this crazy journey

00:56:57.520 we're on. And so that hundred lets us out of it. When you get to the higher material and you've done

00:57:03.220 90, our app's still going to go back and fill those 10% because it's the best way to learn,

00:57:08.300 but it's just get them advanced, let them go and do that. But the hundred, the kids all just do it

00:57:12.520 because they know they can. And it just gets the right mindset going, which is why our kids literally

00:57:18.120 for the SAT, everybody thinks, oh, you can't get perfect scores. They're like, it's just some extra

00:57:21.820 work. Fives on APs, people are like, that seems impossible. They're like, oh, how many hours is it?

00:57:26.380 It takes 75 hours to go through AP world history or AP US history. It's 75 hours. And they're like,

00:57:32.500 oh, I got 15 hours left to get a five. So you just change the concept.

00:57:36.240 Now there's some upper bound where there is conceptual understanding and everybody can't

00:57:40.840 get there, but K through eight curriculum, what would change education in America? We can totally

00:57:46.100 do this. So let's now talk about the AI piece, which we haven't really explicitly talked about.

00:57:51.440 You said something to me once that I think was actually remarkable. You compared it to medicine,

00:57:56.180 you compared education to medicine in terms of an unlock. So talk a little bit about that.

00:57:59.960 If you go back to the sciences and whether it's biology or medicine, chemistry, physics,

00:58:06.420 all of these had periods in the wilderness. I'm sure doctors' bloodletting and things like this,

00:58:12.780 but there was the invention of an instrument in each of those that allowed more precise measurement

00:58:17.780 that unlocked and inflected the science. Medicine had theirs. I don't know which one you thought.

00:58:24.420 I've heard you discuss this. I think it's the light microscope. If you were going to talk about

00:58:28.240 one instrument that took us from what I described as the dark ages of medicine 1.0 into medicine 2.0,

00:58:34.720 which is where we are today, that inflection point came primarily around, without the light

00:58:40.400 microscope, it couldn't have happened. Sid Mukherjee wrote very eloquently about this in his

00:58:44.000 most recent book, The Cell. Basically, until we could actually start to see microscopic organisms,

00:58:50.400 we couldn't really handle germ theory. Until we could eradicate germ theory, we couldn't really

00:58:57.460 live past about 40 on average. There were just too many people dying too young. That one piece of

00:59:04.480 technology unleashed an entire change in the way medicine has been practiced, which of course has

00:59:11.100 spread out far beyond infectious diseases. But I think it's impossible for me to imagine a scenario

00:59:16.620 whereby we are alive today in the way we are without that. When you look at AI, learning science,

00:59:23.140 these papers have been written for 40 years, but they all start with, this doesn't work with a

00:59:27.380 teacher in front of a classroom time-based model. You had the conceptual solution for the last 40 years

00:59:33.220 on how to fix education, but there's been no technology to make it happen. You haven't had your

00:59:38.980 microscope. We haven't had our microscope. And AI is finally that microscope because of a couple of

00:59:43.620 things. The first is, before your effect size, when you put an experiment into a classroom,

00:59:49.420 you're like, well, is the teacher teaching it correctly? You know, is it using space repetition

00:59:53.300 correctly? Are the students engaged? Do they have the prereqs? You just don't know. And so your

00:59:58.460 environmental variables dominate your effect size. With AI, you get two things. You get absolutely

01:00:05.000 precise teaching. You know exactly what is being told to the kid. And then second, you have this

01:00:11.840 closed loop measurement. You can measure what do they know and what they don't know. You know all

01:00:16.440 of it. And then you can create, you can do science. You can create this closed loop. So our system,

01:00:22.740 we rolled out our new system time back to all our students this year, a couple of weeks ago,

01:00:27.820 our math curriculum. We believe this year is going to be better. Kids will learn more in 20% less time.

01:00:35.580 And we know that because we've been measuring it. And we looked on last year's math and we're able to say,

01:00:40.560 okay, we have this idea. Okay, the kids, what percent learned? Were they 95% right? How long

01:00:46.360 did it take each kid to go through the lesson? What do we need to change? What learning science

01:00:50.220 concept can we put in the lesson that's going to then allow them to learn better and faster? And

01:00:55.060 what's going to change the scheme of their brain? Not cramming, but really understanding this.

01:01:00.220 And then, okay, they took the standardized tests. Oh, look, it did work. And you can just create this

01:01:05.300 closed loop system where you're able to do science. Nobody else has that. And that's the

01:01:11.220 problem. We need to create that system. But to do that, you have to give up the most treasured notion

01:01:17.540 in the world, which is the key to good education is a teacher in front of a classroom. One of the

01:01:23.140 things that's true about education, before there was tutors, you had Socrates and Plato and Aristotle

01:01:28.260 and Alexander the Great and all the elite had tutors. And then 200 years ago, like we have to educate

01:01:34.060 everybody. The only way cost-effectively to do that is teacher in front of a classroom.

01:01:38.100 And that is all any of us know, our parents know, our grandparents know. We've all been in that.

01:01:42.860 And it did great things because at scale, we have educated people that never would have been

01:01:48.280 educated before. I wouldn't have been one of the tutored ones. 1.00

01:01:50.760 Correct. It is. And you can go to the poor school district in India, and there's a teacher in front

01:01:54.980 of a classroom, not very nice building. You can go down to whatever private school that started

01:02:00.120 near $50,000, teacher in front of a classroom in a very nice building. But that model is all we know.

01:02:06.020 And great teachers matter. This is always the part with our new model in AI tutors.

01:02:11.380 I want to take a second here because people get freaked out. Adults, 20 years from now,

01:02:16.180 parents are going to drop their kids off at a building. And in that building are going to be

01:02:20.320 other kids and adults. And those adults, you can call them teachers, guide, coaches, whatever you want.

01:02:25.640 Now, we think the day will be very different. But that concept of a school with teachers and

01:02:31.740 other students, that's not going away. We're not moving to this AI robot terminator. That's not

01:02:36.760 what's coming. You're going to drop the kids off at a school. If we do our job right, you don't sit

01:02:40.760 in class for six hours a day and have this miserable experience. Kids love it and all of those. But back to

01:02:46.080 the science of it, we now with AI can have a one-on-one relationship. The AI knows what does the kid

01:02:53.960 know? What do they not know? How do I give them this lesson? And if you look and if you want to talk

01:02:59.060 just about where we're headed, the future is even... Gen AI is changing everything. That was the part for

01:03:06.640 me that I saw, which is I can craft personalized lessons for every kid that takes into account what

01:03:14.720 goes into the LLM, sort of on the technology side. What are you trying to teach them? What's the curriculum?

01:03:18.840 What is their knowledge graph? What do they know and what do they not know? What is their interest

01:03:24.100 graph? What are they interested in? Because actually, if I know your interests, I can engage

01:03:29.880 it. Oh, you want to learn baseball? You love baseball and you want to learn stats? I'm going

01:03:35.720 to terminate in that and kids are going to get engaged. Oh, you're the fashionista? I'm going

01:03:40.240 to teach you fractions. Is the app already able to do what you just described?

01:03:44.160 So where we are today, in 2025, we generate static content with human review because we still have

01:03:50.560 too many hallucinations. We're not there. Now in 2026, we expect we'll be able to do it.

01:03:55.360 Just to make sure I'm putting this in perspective. So my eight-year-old, who is that kid who loves

01:03:59.600 baseball, all he does is talk baseball stats. He is a walking baseball statistic. Do you know so-and-so

01:04:06.620 hit 326 in this many home runs? And do you know he was able to hit 477 feet? Blah, blah, blah, blah.

01:04:12.460 He will be able to get to know his AI tutor. They will figure out in short order that the

01:04:18.540 way to his heart is baseball. And anything that you can teach in baseball terms, he's fixated.

01:04:25.480 It will be able to give him a lesson to pursue mastery, but under the guise of baseball.

01:04:32.600 Under baseball. And we all know this. That's what teachers are always trying to do is how do I

01:04:36.660 get what the kids like and associate with it? But at scale, you don't have 30 kids who like all the

01:04:42.000 same things, et cetera. So here's an example of what you can actually do today. You have

01:04:46.280 hallucinations and you don't want to release those yet. We have a dynamic reading generator. It's

01:04:51.440 called Teach Tales. And the goal of this is to get kids to love to read. So if you take a third grade

01:04:57.220 boy who's like, I hate books. I hate books. I'm not a book guy. And you're like, what's your favorite

01:05:03.020 movie? The Avengers. Who are your soccer buddies? Okay. The AI generates a choose your own adventure

01:05:08.880 where you're saving the world with your soccer buddies. You know, they have to keep clicking.

01:05:12.440 Can you each be an Avenger?

01:05:13.520 Each be one. And literally in the Lexile level, the difficulty goes up, it knows. So it's keeping

01:05:18.580 you in the zone of proximal development within 10% of what you know, your Lexile level. Kids will sit

01:05:23.600 there and click for an hour. Totally engaged.

01:05:26.060 Non-stop reading.

01:05:26.780 Non-stop. Because we do have a reading problem in this country. People don't read books and all

01:05:30.320 these things. This is the kind of thing where Gen AI is going to be able to suck kids in. 1.00

01:05:35.660 Our competition is Fortnite and TikTok. We need to use things like Gen AI to create such

01:05:41.820 compelling content for kids that's related to what they care about. That also teaches,

01:05:47.540 okay, I got a teacher for actions, but we're doing this baseball thing, so you're fine. You love that.

01:05:52.480 There's a whole program that a lot of people in finance where there's actually a program where

01:05:56.360 they have a mother-daughter thing, back to a different way to try to solve it without an AI tutor,

01:06:00.440 where there's all these finance moms who teach their daughters poker in middle school. And they're 1.00

01:06:06.460 like, I'm going to get them into poker to teach all these skills and the math skills and all the

01:06:10.400 things that aren't being taught in my school. But with us, we can get that engagement level where

01:06:16.040 the kids are interested in it. And to give you a couple examples, and I'll use some of our high

01:06:20.480 school kids. So when you transfer into Alpha High, your knowledge of history is very limited,

01:06:27.260 which is if it's in the musical Hamilton, you know it occurred. And if it didn't occur,

01:06:33.580 it's not in a song in Hamilton, you don't know it because we don't teach history very much anymore.

01:06:38.440 But all the kids know Hamilton, they all sing songs. Our US history, AP world history,

01:06:43.480 they literally used AI. They did this on their own. It wasn't our stuff.

01:06:46.420 They used AI to generate, here's all the facts. There's nine units in US history. For each unit,

01:06:53.080 they created a song. And they said, give me all the facts that I need to know for this unit.

01:06:58.580 And they pumped it into Sumo, which is an available song generator. And they created their own album

01:07:04.220 of songs. And they all would just sit around and sing the songs and learn the facts that way.

01:07:09.900 And do it in a way they all were very interested in. And they all got fives on their AP because you

01:07:14.720 can do it to mastery. And so Gen AI is going to allow us to create this awesome dynamic content

01:07:19.880 that's relatable and focused on what, or applicable to what kids want. The other part is there's a

01:07:26.980 learning science concept also bells and frameworks of the fastest way to learn is basically analogy.

01:07:32.080 And so do you have a knowledge graph?

01:07:34.660 This is the A in the ABC. So Dean Schwartz's book, we're going to link to that in the show notes,

01:07:39.820 goes through A to Z, everything that learning science has come up with.

01:07:44.980 Exactly.

01:07:45.380 Right.

01:07:45.760 Dean Schwartz from Stanford, School of Education, one of the leaders in the field,

01:07:49.520 has written a book and he's like, here it is. And every letter has a different one. But this one-

01:07:54.500 A is analogies. One of the ones I remember.

01:07:56.300 Yeah. And analogies, that's the fastest way to learn. So if you know what a kid knows,

01:08:00.220 we have a content team when I talk about those hundred facts, ideas, and concepts per hour.

01:08:03.580 What they're using as reference is they're using TikTok memes because kids know TikTok and they

01:08:10.160 know what's on TikTok. Well, let's take what we're trying to teach them and use something they

01:08:14.900 already know. And that's actually the fastest way to have kids learn concepts. And so those are all

01:08:20.460 coming. Then even when you talk even further, the part we don't have in yet, but our team's working

01:08:25.540 on it is we also take into account, we will be taking into account the cognitive load of their brain,

01:08:30.600 which is we're going to be able to create a lesson where we know how many working memory slots you

01:08:34.980 have. We know what you know to fluency and what you don't. And we're not going to give you a

01:08:39.780 question that's going to overload your working memory. We also know how many reps you need and

01:08:44.880 GT kids need fewer reps. And so we're not going to bore you by giving you too many reps to memorize 1.00

01:08:50.440 it. Conversely, we're going to make sure you have enough reps if you need them. And so this

01:08:55.520 personalized learning where you're generating the content dynamically is going to be just as

01:09:00.460 crazy unlock for these kids. Five years from now, everybody's going to look back and be like,

01:09:05.240 okay, well, obviously kids can learn 10 times faster. We wake up every day. It's like Neo in

01:09:10.260 the matrix. That's a physical upload. Ours is using learning science concepts. Kids can learn 10 times

01:09:16.000 faster. And when you think about that, you have 12 years, you have 12 years to fill your kid's head

01:09:22.660 with cool stuff, knowledge, because they have to hang out in their brain for the rest of their life.

01:09:28.100 And we've allocated 12 years to fill it up. We just need two hours a day at 10 times faster.

01:09:34.120 And all the kids can have more stuff and more cool stuff about stuff they like in their heads

01:09:39.940 to mastery level, levels none of us as parents could ever imagine, and still give the kids their

01:09:46.400 time back for the rest of the day to do all other things.

01:09:50.360 So about three years ago, you sort of decided you got to step out of Trilogy. It's this incredible

01:09:56.220 company you've built. Are any of your co-founders still there? I think it's...

01:09:59.720 No. So three years ago, the transition was when Gen AI came out, which was three years ago,

01:10:06.360 neural nets are here. Finally, that was an unlock for me where I had seen McKinsey's great school

01:10:12.040 and seen my kids in there. And there were issues like my daughter failing her seventh grade tests

01:10:16.660 and stuff where the apps aren't good enough, all those issues. And I'm like, Gen AI will allow us to

01:10:22.100 take the magic of alpha, make it better, and scale it to a billion kids. There's lots of good

01:10:27.480 educational things that just don't scale. And so what this was, was my realization.

01:10:32.200 This was your light microscope.

01:10:33.280 This was it. Light microscopes here. Okay. I've been doing software. I went to my team who've been

01:10:39.180 there for 25 years. I'm like, guys, for 25 years, you've been telling me you're better than me.

01:10:43.820 We'll see. I'm going to go be a principal. And so they run it. I'm now chairman. And I am now

01:10:49.760 literally the worst shareholder you've ever seen. Because I'm like, guys, I need more cash flow.

01:10:54.640 I need more dividends to fund this whole thing. You know, McKinsey needs $300 million to open up

01:10:58.760 all these schools. Where are we going to get it for? So how much of your money have you spent on

01:11:03.360 this now? Three years ago, I took a billion dollars out of Trilogy. It was my first lump sum and said,

01:11:09.980 I'm going to go use this billion dollars to figure out what I can with education and fix as much as I can.

01:11:16.480 Now, what's crazy about that is a billion dollars is not enough to fix education. It's a trillion

01:11:21.460 dollar industry. And we obviously have made incredible headway. But we have to go back to

01:11:26.740 building schools, back to McKinsey. She's opening up 25 alphas around the country. She needs hundreds 1.00

01:11:32.000 of millions of dollars for that. We're opening up these sports academies, which needs hundreds of

01:11:36.040 millions. We're opening up 500 gifted, talented schools. The total capex that's going to be required,

01:11:42.940 the second part is, I'm going to do this the next 20 years, right? I got 20 years in this.

01:11:46.300 My job is to go fix education, get it to a billion kids and give kids their time back.

01:11:52.040 Use the light microscope, use AI to totally change the first 12 years of a billion kids.

01:11:59.260 At what point, Joe, did the two systems have to merge? Because one of the reasons that you're

01:12:03.580 having to deploy such insane amounts of capital is because you're basically rebuilding things that

01:12:10.900 are technically already there. I mean, we have buildings, we have schools, we have guides,

01:12:14.560 they're called teachers. We have all those things, but you're creating a new system of

01:12:18.320 infrastructure. But I assume you don't want to have to build the scale to a billion. At some point,

01:12:23.160 this way has to merge with the existing way. At what point does that merger take place?

01:12:29.120 My view is there's a point of, if I had perfect randomized controlled trials,

01:12:35.480 back to a billion kids, I don't know, I need 10 million kids going through it for everybody to be

01:12:40.160 like, okay, this looks like it works. But the bar is so low. I mean, I guess that's the thing

01:12:44.080 that's amazing to me, right? When I think about places in medicine where massive clinical trials

01:12:49.900 are necessary to change the direction of something, usually when the state of play is so bad, let's

01:12:56.820 talk about an example, which again, I don't use this example to be offensive to people that are

01:13:00.600 within education. But let's look at HIV in the 80s and early 90s. It was a death sentence.

01:13:06.700 The bar was pretty darn low when heart therapy, highly active antiretroviral therapy came out

01:13:14.120 in 95 or 96. And that, boom, we just changed the protocol. Of course, pharma is easier than

01:13:20.260 education in the sense that it's a simple intervention. But who's pushing back when you

01:13:25.700 say, hey guys, what if we just prove this out in a million kids? Why do I need 10 million kids to

01:13:31.580 prove this? Especially if they're spread out over a broad enough socioeconomic background.

01:13:36.120 That's sort of our next phase and why I'm doing my first podcast in 20 years is there's a lot of

01:13:43.740 impediments to it. Let's go back to that discussion of parents don't care about academics. So we'll

01:13:48.980 start there. We'll do that. Second, believability. There's been ed tech pitches. Everybody, every five

01:13:56.280 years, ed tech is going to solve the world. Technology is always going to solve the world and it doesn't

01:14:00.460 solve it. There is good, well-earned natural skepticism of, you know, and let's just go and

01:14:07.840 let's just take it. Evil billionaire tech guys trying to fix education are-

01:14:12.780 It's a meme.

01:14:13.280 No, it's a meme.

01:14:13.920 Yeah.

01:14:14.060 Like make a movie. So the skepticism should be super high and it is. So that's part of what the

01:14:20.340 job is and why it takes a billion dollars. We have to go prove it across everybody and in a broad

01:14:26.140 way. Completely agree with that. But let's talk about some of the problems with it, which is let's

01:14:30.940 talk about academics right here in Austin. There's been three failing middle schools in Austin. And I

01:14:36.020 look at if I was a superintendent of a school or principal of a public school, I mean like, I want

01:14:40.160 to be really clear. That's like a hundred times harder job than mine because I'm a product guy.

01:14:44.780 People who come to my school are aligned with what I'm building and my offering. You get selection

01:14:50.360 effects. People come to Alpha are like, I know I'm going to have screen time for two hours

01:14:53.900 because if you don't, you don't come. But at public school, these three schools are Fs.

01:14:58.640 For five years, they're not educating the kids. And the superintendent, he made his first proposal

01:15:03.120 or whatever months ago. He's like, well, why don't we shut these schools and we have these better

01:15:06.380 schools and the kids just go to the school where the kids are learning. Literally the protests that

01:15:10.880 have been from parents, listen to them. It's the mom who's just like, look, I know that seventh grade

01:15:16.440 math teacher. She's probably not teaching math very well, but my older daughter had her. 1.00

01:15:20.620 She changed her life and my kids can walk to school and it's community. And so if they don't

01:15:26.020 learn seventh grade math so much, so what? The superintendent, if you take into account the

01:15:30.080 community feedback, which they're supposed to, all of a sudden you have parents who are like,

01:15:33.780 I care about other things because school's a bundle. And so that makes this hard. So there's

01:15:37.960 one just, do they even want it? The second is obviously just skepticism of this doesn't really

01:15:42.600 work. And why don't we do randomized control tiles or pharmaceutical caliber stuff with education?

01:15:48.720 Because there's all these ideas of work. We have a lot of ideas that we're rolling out,

01:15:52.980 like no multiplication tables that somebody should have done randomized control tile to realize

01:15:57.540 that's a really bad idea. And so all these fads in education, because it's not science,

01:16:02.480 the learning science, if you actually ask one of the scariest data points, if you want this,

01:16:07.660 there's 3 million teachers in the world. They're not taught learning science. They're not taught

01:16:11.480 learning science. These concepts will be new to them. And so back to how do we fix it? We have to go

01:16:17.840 teach all the teachers about learning science. Do you get pushback on that point? I mean, how much

01:16:22.340 time do you spend talking with the teachers that are in the trenches right now? I'll give you an

01:16:26.580 analogy from my world. I talk about this idea of medicine 3.0. Medicine 2.0 has been amazing.

01:16:31.620 It's an amazing run. You and I wouldn't be alive without it. Literally, we would be dead by now.

01:16:36.660 But I would argue it's time to move on. We've plateaued. We got to do the medicine 3.0. We have to

01:16:42.180 unlock the next thing. But when I say that, I'm not being critical of the doctors who practice

01:16:46.380 medicine 2.0, because I haven't met many of them who wouldn't in a heartbeat say, look,

01:16:50.900 I wish I could do medicine 3.0, but the system I'm in doesn't allow it. I can't get reimbursed

01:16:56.280 to practice medicine 3.0. The entire system of billing and coding and diagnostics and prescribing,

01:17:02.840 it's all predicated on this system. It's predicated on make the diagnosis for the disease,

01:17:08.980 treat the disease. There's no prevention. I don't get paid to, nor do I have the education to learn

01:17:13.740 about nutrition, exercise, all of these other things that are necessary to medicine 3.0.

01:17:18.800 But if I could wave a magic wand, my bet is 80% of them would be like, if there's a way to fix the

01:17:24.960 system and the structure such that I can be balancing my practice between 2.0 and 3.0,

01:17:29.640 I do in a heartbeat. Do you have a sense? Have you got enough reps with teachers to understand if you

01:17:33.620 had that proverbial magic wand, how many of the superintendents and the teachers would say,

01:17:37.880 oh my God, I'd love to transition from being a front of the classroom teacher to an in the

01:17:44.200 classroom guide who's helping the student with their personal AI tutor?

01:17:49.340 So the answer is, it is at least 80%. Like same example, which is-

01:17:52.600 So you're on the same team?

01:17:53.660 No, no. Because we have our guides, they're ex-teachers. We got into this whole naming thing,

01:17:59.220 but do you call them teachers if they're not doing academics and whatnot? But fundamentally,

01:18:03.320 when they tell their story of what they do all day, and let's just talk about teachers specifically,

01:18:07.700 teachers are great and critical and teacher in front of the classroom is bad. But to your point,

01:18:12.580 it's the system that-

01:18:13.480 It's a very similar analogy.

01:18:14.700 It's the same thing. Teachers, they got into teaching to transform kids' lives. They did not

01:18:21.340 become a teacher to grade the seventh grade science test. And so let the AI do the seventh grade lesson

01:18:27.620 and the science test and you connect with your kid, your students. So during our two-hour

01:18:33.080 blocks, we do 25-minute Pomodoros. Our guides, ex-teachers, they're not in front of the class

01:18:39.100 lecturing. They literally take the kids one-on-one and say, hey, don't do your language today. Come

01:18:46.880 sit down and talk to me for 25 minutes. How was your weekend? How'd the softball tournament go?

01:18:52.060 Hey, you don't seem as motivated lately. What's going on? And they connect with them. Our guides,

01:18:57.240 they provide motivation and emotional support. They set high standards and high support. They're the

01:19:02.220 ones guiding these kids through the journey. And that's why teachers became teachers.

01:19:08.280 I want to write a lesson plan. Nobody woke up and said, that's why I'm becoming a teacher. They did

01:19:11.780 it because they want to transform kids' lives. And that's what this new model does. And so when they

01:19:16.480 see it, they love it. So one, we pay more, but also the teachers see our model. And once they really

01:19:21.860 understand it and talk to our existing ones, they're like, this is awesome. I love this.

01:19:26.280 And so there isn't resistance. Now at the systems level, so I've talked to a lot of superintendents

01:19:32.580 and they're like, what do you want me to do with this? What am I supposed to do? They're like,

01:19:37.940 what do the kids do the rest of the day? The problem when you talk about it, this is a rebuild

01:19:42.380 around the school day. It's not six hours a day in class with teachers and removing kids between

01:19:47.980 two hours a day of learning in the screen and four hours a day of life skills and sports and all the

01:19:54.980 sorts of other things that are going to be imperative to create well-balanced kids.

01:19:58.980 Right. And our life skills curriculum is leadership, teamwork, storytelling, public speaking,

01:20:05.320 relationship building, socialization, grit and hard work, entrepreneurship, financial literacy,

01:20:11.840 those things. And they're doing workshops in the afternoon. This is if you go to McKinsey's

01:20:15.680 Instagram, kids climbing rock walls in kindergarten, or they're running five K's in second grade,

01:20:21.420 or the fifth graders launched a food truck. So there's all these different things, running

01:20:25.920 Tough Mudders, eighth graders do that stuff. If you're a superintendent, you're like, okay,

01:20:30.140 my system's not set up for this. And right now there's no demand from the parents because it's

01:20:35.460 not proven. It's not obvious. And let's even get out of public school. Let's go just talk about

01:20:39.980 super high-end private schools. You can go down the street and try to convince a super high-end

01:20:44.340 private school and say, you have all the flexibility you want. There's no vote. There's

01:20:48.200 nothing. You should transform. And they're like, this seems like a bridge too far. And so that's

01:20:54.320 my view of my job. So is that why essentially Alpha exists as an independent school at the moment

01:20:59.860 is you just need to get the scale of wholly owned Alphas that build this out. And at some point,

01:21:07.400 the others, probably the private schools have an easier time joining the model. And eventually the hope

01:21:14.020 is of course the public schools do so as well. Everybody. So I got 20 years and I got to get

01:21:18.700 a billion kids. 80, 90% of kids are in public school. You got to get it to public school and

01:21:22.780 provide it to everybody. What is the current cost? Because I'm sure you're hemorrhaging and losing

01:21:28.040 money at the moment. What is the current cost to deliver the education to a child? And does it

01:21:33.760 vary by their grade? So stepping back, so for Alpha, our job is, and that's why we're opening across

01:21:38.620 country. In every city, there's going to be this Alpha exemplar, the high end where you're like,

01:21:43.800 this is what's possible. Now it's also one where it's money, no object. It's really expensive.

01:21:48.820 And back to the things today, my AI, I'm using the latest models, this real-time streaming video

01:21:54.540 to coach kids. It's 10,000 bucks a kid.

01:21:56.640 $10,000 per kid per year in just AI cost.

01:21:59.520 In just the AI cost. It's too much. Now, I absolutely believe, and our team's going to

01:22:04.120 deliver it, where sometime in the next three years, maybe five, everybody on the planet will

01:22:09.300 have a tablet that's less than $1,000 that will teach them everything they need to know in two

01:22:15.160 hours a day or less.

01:22:16.140 And the compute will be what?

01:22:17.680 It'll be local.

01:22:18.440 It'll be local.

01:22:19.340 So you'll be on device.

01:22:20.600 Wow.

01:22:20.720 Like a lot of our stuff right now is vision models.

01:22:22.560 We're looking stuff. There's on device, all the hardware manufacturers are putting their

01:22:27.860 AI chips on device. We'll hit that. I'm sure there's still be some stuff in the cloud for

01:22:31.840 higher end stuff, but the common core curriculum through eighth grade on device, less than $1,000.

01:22:36.800 That's absolutely coming. Then the second part is, part of what Alpha is, is because it's this

01:22:42.640 high end and expensive stuff, you're like, let's re-envision what a school is from the ground

01:22:47.420 up. So we sit around, I'm like, what would make a school 10 times better than Old

01:22:52.540 School that we went to? We sort of have five dimensions that we do. And the first is,

01:22:57.640 kids must love school. And we literally measure more than vacation. And we get 50-ish percent of

01:23:02.560 kids, depends on the vacation, depends on the workshop if it's really good, 50% of kids who are

01:23:06.740 like, yeah, I want to go to school instead of vacation. My highlight as a principal was in May

01:23:12.640 this year, two-thirds of the high school kids sent me a note and they're like, can we keep Alpha

01:23:18.280 High open this summer? Because we don't want to take a summer break because we love what we're

01:23:22.140 doing. The girl doing her Broadway musical, she's like, I want to keep coming to school and doing

01:23:26.180 it. And so back to my world where I was not into school and I was shirking and my dad, I'm like,

01:23:31.760 make a place where kids love school and vacation's a good standard. Number two, kids need to learn 10

01:23:36.820 times faster. We need an engine that teaches kids 10 times faster. And so our AI tutor based on learning

01:23:42.000 science teaches kids 10 times faster between five and 10 and we'll get it up. So that's the second

01:23:46.780 part. Then that gives you the chance to give the time back. The third dimension of a school

01:23:51.220 is academics isn't everything to developing kids. Everybody knows that. What are the life

01:23:55.900 skills that every parent says, that is what I wish I knew. And some are traits, behavioral

01:24:01.060 traits. I hope they are love learning and are self-driven learner. You want things like that.

01:24:06.880 And I want somebody who uplifts others in his classroom and behavioral and is an independent

01:24:11.160 person. Others are like those ones I talked about where you're like, no, I want entrepreneurism and

01:24:17.000 financial literacy, or I want storytelling, public speaking, those type skills. And so the third is

01:24:22.540 have that curriculum. The suite of skills.

01:24:24.840 The suite of skills that every parent wants. So we actually have spent the last three years

01:24:29.300 developing LifeCore, the common core curriculum for life skills, because I have four hours a day to

01:24:34.380 fill with all this awesome stuff that kids can do. And it's just amazing what they do. If you take

01:24:39.220 Atomic Habits, that book of 1% better, most people don't read that till after college.

01:24:45.300 And then they're like, oh, this is a way I can really develop. Last year, our guide, Faith,

01:24:51.120 second grade guide, she sat down and she was working on her workshops, you know, her life skills. She's

01:24:56.000 like, you know what? I want to teach Atomic Habits to all the second graders because the concepts they

01:24:59.960 all can understand. And she's like, you know how I'm going to do it? She's like, I'm going to have

01:25:04.260 all our second graders run a 5K in under 35 minutes. I'm like, oh no. Okay. The parents

01:25:10.880 are not going to like that. That's too hard. Back to high standards. They don't want struggle,

01:25:15.420 fail. Where's that road to success part and what percent are going to make it? She's like,

01:25:19.600 don't worry. I can do this. She was a UT athlete. And she's like, we totally can do it. She was like

01:25:23.740 a head cheerleader or whatever. She's like, everybody can do this. So she sits down with the 0.99

01:25:28.600 second graders first day and is like, who can run a 5K? And they're all like, that's impossible,

01:25:32.780 miss faith. The parents all are like, that's impossible, miss faith. She's like, well,

01:25:36.460 I signed you all up for the jingle 5K, you know, at the domain. So she goes out to the track in the

01:25:41.180 first day, they walk it and they walk 5K. She's like, okay, so we all have established you're going

01:25:45.600 to finish the race. And then the next time they ran a quarter lap, walk the rest. Next time,

01:25:52.000 half a lap, walk the rest. And by December, all the kids are running the race. And, you know,

01:25:58.320 when they cross the finish line, just that achievement, back to the key to your kid's happiness is

01:26:02.580 high standards. That's a really high standard, but they loved it. And they did have to struggle

01:26:06.900 and go through and learn this, but they also learned that life skill of 1% better. And so

01:26:12.880 the second graders like, dad, I can do anything. I can do 50 pushups. I'll do one today and two

01:26:18.040 tomorrow and build up. And that skill, that life skill, we don't teach kids in school. It's a great

01:26:23.420 life skill to have. And if you have half the day, you can do that. So that's the third element of like,

01:26:28.000 how do you build a great school? The fourth element, and this gets to the teachers,

01:26:32.000 what would make a teacher 10 times better than what it is today? And the answer is, well,

01:26:37.140 not grading science tests all day and writing lesson plans, but instead connecting with the kids.

01:26:43.780 Here's the questions we ask students and the parents, which is to kids, kindergarten, we're like,

01:26:49.180 do you love your guide? But, you know, as you get older, it's every adult had one or two teachers who

01:26:55.040 transform their life, right? Back to that. Is your guide that for you? That is the expectation

01:27:00.580 that they are the ones that the students are like, you are helping me achieve things that I

01:27:05.460 didn't think is possible. Ms. Faith, you taught me how to do the impossible. That level. And actually,

01:27:11.440 then the parent side, this gets into Dr. Yeager has a book, 10 to 25, child psychologist,

01:27:18.380 development expert. And it's what matters for kids from 10 years old to 25. His whole theory,

01:27:25.840 mentor mindset, high standards, high support. Most parents are not high standards and high support,

01:27:31.740 which is what really matters to develop kids. They're either low standards, high support. That's

01:27:36.720 what happened to all our expectations. Or the inverse, high standards, no support. This would be

01:27:41.060 your classic tiger mom. Good luck. Figured out. And the real answer for child development, you need both. 0.96

01:27:46.640 High standards, high support. And so that's our guide. But the question we ask parents in middle

01:27:51.040 and high school, back to having what's a school, re-envisioning it, it's like, do you trust your

01:27:56.320 alpha guide to hold high standards for your kid so you as the parent can provide the unconditional love

01:28:04.620 and support? Which transforms, for those who've had adolescence, it transforms your relationship where

01:28:10.880 you don't have to be the naggy parent for the high standards. My oldest daughter, she's writing her 1.00

01:28:16.120 college essay. And I'm like, you want to let me read it? And she's like, no, no interest in that.

01:28:22.080 She's got it taken care of. She's got someone else. 0.99

01:28:23.640 And for me, I was super relaxed because I was like, I know Chloe, back to our guide. She's a

01:28:29.300 Harvard grad. I know she was totally high standards for my daughter. And I know it's going to be fine.

01:28:34.020 And I don't have to get in a fight about it or whatever. I'm just like, okay, yeah, I can give my

01:28:37.340 daughter a hug and be like, good luck. That transforms what teachers can be versus I'm

01:28:44.140 delivering those seventh grade lessons. And so the fifth dimension that we talk about is the C's,

01:28:49.200 character, community, classmates, and culture. And these are also the hardest parts to measure,

01:28:54.760 but at the end, they're what parents care about the most, which is at the end of the day,

01:28:58.280 you don't care. Do I have a good kid? Did I raise a good kid? And is the school helping me

01:29:03.480 raise a good kid? Those elements are what we looked at and said, now that you want to re-envision

01:29:09.680 school from the ground up, here's the dimensions. So alpha is trying to be the exemplar for all of

01:29:16.600 it. And then the question is, okay, how do we get the price point down? So we have opened a whole set

01:29:22.720 of different schools at half the price of alpha. So alpha, depending on its location, is between

01:29:27.720 $40,000 and $75,000. Although there's scholarship programs.

01:29:31.160 Tons. So we have a huge scholarship. We have a school down in Brownsville where it's the second

01:29:36.060 poorest school district.

01:29:37.520 In Texas?

01:29:38.200 In the country. Brownsville is the second poorest school district in the country. At that school,

01:29:42.060 it's SpaceX is there. It's so star-based. So you have a bunch of SpaceX kids. And then we scholarship

01:29:47.300 all the locals to come in and it's 60, 40, 50, 50 between that makeup. And to be clear,

01:29:54.360 they all learn 2X in two hours. Back to testing it out. It works with all kids.

01:29:59.380 So I want to read you something that fits right into this. So this is, I think I read

01:30:03.920 this in the New York Times. Before kindergarten, the children of the affluent are much more likely

01:30:09.620 to be in preschool. By sixth grade, students in the richest school districts are four grade

01:30:15.460 levels above children in the poorest school districts. That's just by sixth grade. By high

01:30:20.720 school, richer kids' average reading skills are five years ahead of poorer kids. By college,

01:30:27.800 according to a 2017 study by Raj Chetty, children from the richest 1% of earners are 77 times more

01:30:36.620 likely to go to Ivy League schools than children from families making $30,000 a year or less.

01:30:43.200 The academic gap between the affluent and less affluent is greater today than the achievement

01:30:51.240 gap between white American and black American kids in the final days of Jim Crow. In other words,

01:30:58.060 my read of this is that there is no greater predictor of academic success and failure

01:31:03.400 than the wealth of your parents. So are you really suggesting that that could be wiped out?

01:31:10.960 There's always relative versus absolute. And I believe we can raise the absolute standard.

01:31:18.660 It's always an arms race.

01:31:19.780 Put it this way. Instead of the gap being 77x, could that gap be 1.2x?

01:31:26.360 Yeah, 100%. 100%. Back to that issue, because here's what happens. You're in a rich school district.

01:31:32.000 I was up in New York and these are $75,000 private schools, teacher in front of a classroom. Every one

01:31:38.080 of those parents also was spending $750 an hour on a tutor after school, because the school wasn't

01:31:46.240 teaching them, back to that problem. But an extra 75 grand tutoring them, 150 grand a year. Yeah,

01:31:53.060 you need the tutor who's going to go back and fill the holes and stuff. And no, if I am in a school

01:31:58.220 district, poor school district. I don't have the tutor. I don't have the access to that. If you look

01:32:02.800 after COVID, the number one intervention, a lot of government funding was spent this was high dosage

01:32:08.580 tutoring. Everybody knows this. It's not a secret. It's expensive to give individualized tutoring. And

01:32:15.340 there was a study that came out a few weeks ago or a month ago that was like, it was really working.

01:32:19.920 And then sometimes as it expanded out, it did less well. And the difference was when it was really

01:32:25.140 working, it was one-on-one and then they started to save money. They had six to one and tutoring six

01:32:29.800 to one is because it's not individualized. It becomes a class because it's just cost-driven.

01:32:34.420 That's the issue with, if you have 30 kids, you can't generate an unending stream of content at

01:32:40.360 80 to 85% accuracy based on what the kid learns. It has to be one-on-one. That is the delta for each

01:32:46.000 kid. It's personalized learning. A personalized lesson plan is what every kid needs. And I believe that

01:32:52.020 if you get all the learning scientists together and say, yes, AI is the light microscope, don't you

01:32:58.080 agree? And can we create a program that's available to everybody? And I think they'd all say, yes,

01:33:04.320 yes, we can. And the software is not insurmountable. I was actually talking to Stanford has their Stanford

01:33:12.240 accelerator for learning and Dean Schwartz runs it. And I was like, you know, we could probably create

01:33:16.700 an XPRIZE for an open source available to everybody. You also need better credible people

01:33:23.320 than me. You don't need me saying this. You need Dean Schwartz and Stanford and the Ivy League,

01:33:30.180 the people that everybody looks up to in education in America and saying, we know this. We know this

01:33:35.440 learning science. We can run. We can take Stanford CS and AI department with the learning science

01:33:41.200 and put that together and create this project for not that much money that we can then get to every

01:33:47.240 kid, get it into every public school. They have the relationships with all the teachers where

01:33:52.240 they don't need the randomized controlled trial or the skepticism you'd obviously get from talking to

01:33:56.600 me. So this is all very doable. That is this mission. What is the biggest risk right now? Is it

01:34:02.680 the equivalent of what we call market risk where it's just adoption? It's just the leap of faith that

01:34:08.620 we can somehow abandon the only model any of us know, or is there still technology risk embedded

01:34:16.060 within the AI? There is still some technology risk, but that's literally like you can ratchet down and

01:34:22.960 you don't have to be 10 times learner, but could we get them three times, five times today with more

01:34:26.900 static versus some of the cool dynamic stuff? Yes, you could. But it's, are you willing to change the

01:34:32.100 motivational model for the kids? Are you willing to tell the kid, use this app and I'll give you your

01:34:37.140 time back? Are you willing to say you don't have to sit in class for six hours a day? It's the rebuild

01:34:41.940 that's the problem. If you take our software, our magic AI tutor, whatever it is, and you put it in a

01:34:48.020 standard classroom and the kid has to sit there for six hours a day, I can promise you it won't work.

01:34:53.120 It will not work. And so that's the issue is back to what Brian told me a decade ago. Are you really

01:34:59.940 willing to say kids must love school? Are you really willing and then go rebuild your school system

01:35:04.280 around that? Because if they do, they're going to engage in the app. You're going to give them their

01:35:08.020 time back to do stuff they love. Is that all, are you going to make those changes? Because it isn't

01:35:12.560 an ed tech solution. It's not go build this ed tech, drop it, and it's going to work. We have that.

01:35:17.920 Those apps do generally work. To get the real revolution, you need to say, are we going to rebuild

01:35:23.500 what we think K through 12, the first 12 years of the kids' lives are? And if that's true, then the sky is

01:35:30.720 the limit. And what we're obviously doing is trying to say, look at this, look at these examples.

01:35:36.960 Because parents need, I've learned as a principal, they need two things to want to change the school.

01:35:42.020 Remember, I was skeptical for two years. I'm the average parent. They need a reference from an

01:35:46.940 adult they trust. How's it working for your kid? Did it work for your kid? And second, they need to

01:35:52.820 see the kid do something at the school that their kid can't do. And if you have that, that's actually the

01:35:57.720 magic that unlocks a parent saying, okay, I'm going to go move to this new model. What it is that the

01:36:04.100 kid can't, oh, he loves school. Oh my God, he wakes up in the morning. He's waking me up so he's not

01:36:07.540 late. And he used to cry on the way to school. Or, oh, wow, some parents are like, I love the academic

01:36:12.300 performance. Or those life skills I really care about. Or he connects with his teacher so much

01:36:17.000 more as guide. Because it is different. Back to the choices, like why we were opening all these

01:36:21.380 different schools. We have a gifted, talented school. We have a sports academy school. We have a

01:36:26.420 wilderness school. Fishing and archery and farming. We're going to be opening arts, theater, music.

01:36:31.700 Parents want different things and kids want different things. And so it's showing these

01:36:36.780 things that is going to be the connection that causes parents to say, I want this. And if you do

01:36:41.540 a good example about motivation, if you put the GT moms, McKenzie has her roundtables, 1.00

01:36:47.340 the GT moms with sports academy moms.

01:36:49.160 Gifted and talented as well.

01:36:50.660 Sorry, that's GT. Gifted and talented. And then the sports academy moms, they just look at each other like 1.00

01:36:55.880 the other parent is off the rocker. Where the sports academy mom's like, my kid just comes home

01:37:01.680 just sweating, running around all day. And I literally throw them in the shower with their

01:37:05.160 clothes on. And the GT moms are like, okay. They were having a chess tournament in the afternoon.

01:37:10.160 And then we came home and continued it. And it depends on the motivation. Each kid's different.

01:37:14.540 And all your kids are different. And everybody knows that. And when we ask, like I sat down with

01:37:19.400 the guides, we're like, okay, how do we make these all better? So on staff days, because we have

01:37:24.660 all the guides from all the different schools. And I'm like, okay, what's going to make kids really

01:37:28.140 love school? How do we get it from 50 to 60 to 70 on love school? The GT guides are like more,

01:37:34.140 they want a third power hour of academics and they want math Olympiad. And all the other guides are

01:37:39.620 groaning because they're like, no, our students do not want more academics. They want sports or let's

01:37:44.380 do a better workshop or what it's going to be. And so the key on all of this is saying, how do we

01:37:49.520 focus school on making it awesome for kids, making it great, have a culture where it is high standards,

01:37:56.160 high support, where you have this magic unlock of they can learn 10 times faster, which then

01:38:00.720 enables you to give them the time back to do all these other things. But that's a big lift. That's

01:38:06.280 why it's going to take 20 years. But I do believe that every parent, the most important thing is

01:38:11.880 raising your kids. It's the same at a societal level. The most important thing a society does is raise

01:38:17.560 its next generation and everybody cares about it. We spend enough money in this country or globally.

01:38:24.060 There's enough money to actually rearrange this and build an education system that's great for the

01:38:30.780 kids. 5% of GNP, $7 trillion. There is recognition at both the societal or parent level that this is

01:38:39.580 where we should spend money and we do.

01:38:42.000 Joe, it's kind of amazing to think back in the spring of 1989, you drop out of college.

01:38:47.560 To go start a software company, obviously there's no way you could have imagined both the success you

01:38:53.140 would have in that company, creating a multi-billion dollar company and along with it a fortune for

01:38:57.920 yourself, but then somehow deciding right at your peak that you were going to completely walk away from

01:39:04.520 that, take that fortune and pour it into another endeavor that seems as far from anything as possible

01:39:10.840 relative to what you started with. Look, I feel really fortunate that you started this in Austin.

01:39:15.180 For the listener, I think I probably should have said this at the outset, you know, my daughter

01:39:18.580 spent two years at Alpha. It was an absolutely exceptional experience for her. Really experienced

01:39:24.460 all the things you talk about. So showed up in, what did she show up? She showed up in seventh grade,

01:39:29.460 shows up in seventh grade with the absolute negative story of, I love reading. I love arts. I love

01:39:36.920 these things. I'm a musician. I just don't do math and science. I know you did dad. That's not me.

01:39:42.320 I can't do math. I never want to do math. I made the awful mistake of spending the past year trying

01:39:47.780 to help her do math. That was a disaster that only resulted in tears. But really, I think what appealed

01:39:55.400 to me when we made that switch was it just seemed so entirely logical that the reason she was struggling

01:40:03.980 was because there were things in third and fourth grade she never mastered. And I really appreciated

01:40:10.540 the compounding nature of that gap just made sense. It really clicked. And so it wasn't a big leap of

01:40:17.480 faith to say, oh, okay. Yeah. And I feel very fortunate that just as you predicted by the end

01:40:24.240 of eighth grade, I mean, she was acing math and that has continued all through high school. She did go

01:40:29.160 back to public high school. I think that's probably something you do encounter is there are kids that

01:40:33.820 ultimately just want to be back in the big public high school. They want to play sports in the public

01:40:38.460 high school. They want to be, you know, and of course she's at a high school with a billion kids

01:40:42.080 and she's very happy there, but I don't think there's a chance. I don't think there's a snowball's

01:40:46.500 chance in hell she would be doing as well as she is if not for the fact that she had that two-year 1.00

01:40:51.680 sabbatical to Alpha where she really got to academically create a solid foundation.

01:40:59.880 And I will share one last anecdote here, which is at her eighth grade graduation, she was in tears

01:41:07.540 saying goodbye to her guides. To your point, think of what the impact of those guides were on her life.

01:41:13.720 She loved those people. It's not that she doesn't like her teachers in high school, but she doesn't

01:41:18.500 have that kind of relationship with them. Right. I came in actually right then and Andy M who runs

01:41:25.680 her academics since she brought her up, wants to give her a call out because she spent hours with

01:41:31.400 him going through lessons, helping figure out why is this not working or what is it? And one of the

01:41:39.200 things when you're starting these new schools, you need this concept of a founding family. Your family

01:41:44.700 was one of the most important and your support, you got it. But your daughter, Andy wanted to say

01:41:50.600 thank you to her because she did go above and beyond where our lessons and all this learning

01:41:56.860 science in this closed loop. He's like, I'm way better because of the feedback and the time she

01:42:01.340 spent with us. It is transformative to kids. And to your point of high school, if you want to be the

01:42:07.640 quarterback on the Westlake team and run track, we don't have those in our high school. And so we can

01:42:11.280 get you prepared. We say our middle school prepares kids for any high school. So all our kids in middle

01:42:16.940 school can get their pick of high school that they want to go. And some want alpha, you know,

01:42:21.260 alpha high, which is that, but some want the traditional play, but we hopefully have equipped

01:42:26.080 them where they know they're a self-driven learner and can do it on their own.

01:42:29.920 Yeah. I'll tell you maybe the last story I'll share, Joe, which is no surprise to you is

01:42:33.640 it really doesn't matter what corner. And I mean this literally what corner of this planet I am standing

01:42:39.300 on. When education comes up, your name always comes up. And these are people who don't know that I know

01:42:45.200 you. They don't know that we are friends. I will be literally on the opposite side of this planet.

01:42:50.020 And if I'm talking with anybody about education, it's like, what do you know about AI and education?

01:42:54.840 Have you heard of this guy, Joe Lehman at alpha? And it comes up so often, which tells me that it is

01:43:00.920 getting out there. This is no longer just that little niche school, downtown Austin. I mean,

01:43:06.000 it's hard to believe what has happened in five years. It went from one building, one place to how many

01:43:12.720 now? We opened 12 campuses. This will be a 25 alpha campuses. We also have 25 non-alpha campuses

01:43:20.080 where back to lower price points. We have campuses open where it's charter. So it's free for kids.

01:43:26.000 We have ones where it's $5,000 pay for the parents, $500 a month, some at 20,000. So we're trying to

01:43:32.400 figure out back to innovating. We're like, what are the right price points and everything? But yes,

01:43:36.400 we expect 12 months from now, there to be hundreds and hundreds of schools that use this model.

01:43:43.760 The demand for the sports academy is through the roof. And we think that's going to explode for

01:43:48.780 sure. Well, Joe, I appreciate you taking your time to come out today, but of course, more than anything,

01:43:52.860 I just appreciate what you're doing. And again, I know this is not something that necessarily fits

01:43:57.100 into what we talk about on this podcast, but again, I don't think you can care about medicine,

01:44:01.920 science, health, and not care about the foundation upon which it's built. And it all starts with

01:44:06.800 education. So you're working on such an important problem. And if there's one last thing I can do,

01:44:10.980 since I've had a great career, obviously before this, and three years ago, moved to education.

01:44:14.740 If there's other people out there, and one of the things, if you're listening to this

01:44:18.280 and you're excited, part of my job is let's go get more talent into education. And I can tell you

01:44:25.260 the last three years have been awesome. And my next 20 years is going to be better than my last 20 years.

01:44:30.620 If I can convince you to come on in the water. What's the best way for, because I guarantee you

01:44:35.060 there's a million people listening to us, a non-zero number of them are going to want to

01:44:38.800 reach out to you. What's the best way to get ahold of you? Literally my email, which is

01:44:42.260 joe.lemont at alpha.school. Send it to me. I am overloaded right up to your point. The hype has

01:44:49.320 gotten very high about alpha. What could someone who wants to get your attention put in the subject line

01:44:53.900 heard you on the drive podcast or something on the drive podcast. And we're looking for all sorts of

01:44:59.700 people. You want product people. You want engineers. You want everybody who wants to come in. We have

01:45:04.000 to go build AIs. Back to expansion, we need entrepreneurs where GT school needs 500 schools.

01:45:09.820 We need entrepreneurs who are going to go expand it like it's a Chipotle chain. We need AI guys

01:45:15.460 going out to Silicon Valley where current chatbots aren't good. We have to make the LLMs better.

01:45:20.640 We need educators. People who are like, okay, I want to do this new thing. If you're one of the

01:45:25.780 learning scientists that we don't have on our team, come do it. We're going to need a lot of capital

01:45:30.520 as well to expand this. Philanthropy, right? It's we need to be giving scholarships. We're funding a lot

01:45:36.640 of it, but we need to make this available to everybody. It's your problem. It can't just be

01:45:40.800 for the elite. It's got to be for a billion kids. So education touches everything. And if you're just

01:45:46.280 in where you're like that description of what they're building, I'm in, then come aboard because it's

01:45:52.940 awesome. Joe.Liemont at alpha.school. Okay. We'll put that in the show notes just so that people can

01:45:57.760 click directly on it. Joe, thanks again. Thank you very much. This is great. Thank you for listening

01:46:03.300 to this week's episode of The Drive. Head over to peteratiamd.com forward slash show notes. If you

01:46:10.740 want to dig deeper into this episode, you can also find me on YouTube, Instagram, and Twitter,

01:46:16.320 all with the handle peteratiamd. You can also leave us review on Apple podcasts or whatever podcast

01:46:22.760 player you use. This podcast is for general informational purposes only, and does not

01:46:28.120 constitute the practice of medicine, nursing, or other professional healthcare services,

01:46:32.060 including the giving of medical advice. No doctor patient relationship is formed. The use of this

01:46:38.100 information and the materials linked to this podcast is at the user's own risk. The content on

01:46:44.060 this podcast is not intended to be a substitute for professional medical advice, diagnosis, or treatment.

01:46:49.120 Users should not disregard or delay in obtaining medical advice from any medical condition they have,

01:46:54.960 and they should seek the assistance of their healthcare professionals for any such conditions.

01:46:59.720 Finally, I take all conflicts of interest very seriously. For all of my disclosures and the

01:47:04.820 companies I invest in or advise, please visit peteratiamd.com forward slash about where I keep an up-to-date

01:47:12.780 and active list of all disclosures.

The Peter Attia Drive - September 29, 2025

#366 ‒ Transforming education with AI and an individualized, mastery-based education model ｜ Joe Liemandt

Episode Stats

Summary

Transcript