Superschools for Every Kid

Our kids deserve better

We all agree educating our kids is monumentally important.

Yet school has barely changed in 200 years.

We went from steam trains to reusable rocket ships… but classrooms are still one adult talking while 25 kids try to sit still.

Picture an alternate reality where schooling improves as quickly as smartphones.

Kids would graduate at age 18 with the natural love of learning they had at age 5. Along with confidence, strong reading/math abilities, and most importantly… the belief they can improve the world by building a business, pushing forward science and technology, or helping people in another way.

What a world, right? Imagine all the wealth, jobs and fulfilled young humans we’d have.

For the first time ever, innovations in education are putting this world within reach. I met some of the incandescently bright kids who are taking advantage of these innovations, as well as the founders pushing them forward.

We have a lot to be rationally optimistic about in education!

Just so you’re aware, I’m biased. I believe AI, used correctly, can make school so much better. I want innovation to happen quickly, so my kids, age 7 and 4, benefit. My family has experimented with several do-it-yourself AI-education enhancers that I’ll tell you about in “Parents’ Corner” near the end of this Deep Dive.

Before we look to the near future of education, let’s look at the present.

The Education Depression

Schools are really struggling lately. Test scores are backsliding. More American 8^th graders than ever before can barely read.

Most people assume that’s the fault of COVID’s way-too-long school closures. But that only exacerbated the problem. Basic measures like math proficiency and reading comprehension peaked in 2013 and have been declining since 2018.

Tim Daly, who writes the excellent Education Daly, calls it the “education depression.”

Look at how 8th-grade math performance dramatically improved from 1990 to 2013… but has only gotten worse since, across all racial groups. This charts the US, but Europe’s educational decline looks similar:

Why did schools suddenly start backsliding around 2013? We can find a big clue in the differences between US states, which have a fair amount of leeway to run their education systems differently.

It turns out that several states with formerly world-class school systems have plunged in quality. In the 1990s, Maine students consistently ranked 1^st or 2^nd in reading and math. By 2022, their average rank had cratered to 36^th.

Vermont has suffered the steepest decline in reading scores of all 50 states, despite spending the second-most per student. To give you a dad-on-the-ground perspective, I moved my family to Vermont in 2017 largely because of its excellent schools. We are moving out this summer before my youngest enters kindergarten.

The performance of these and other northeastern states is so bad the Boston Globe named it the “New England Nosedive.”

Meanwhile down south, Mississippi is skyrocketing up the rankings. It has surpassed Maine in 4^th-grade reading scores:

For my non-American friends, Mississippi is the poorest US state. It has the highest rate of single-parent households, which tends to weigh on school outcomes. And it spends less than half per student ($12,394) vs. Vermont ($26,974).

Education experts are calling this the…

Mississippi Miracle

And it’s not just Mississippi. Louisiana and Alabama are also surging past the north in school quality.

You’re probably wondering how Maine and Vermont squandered their formerly excellent school systems. I will leave you to explore this question yourself, as the answers are politically charged. I’ll just note they have embraced “experimental” methods of teaching that sound nice and caring, but clearly do not work. Hopefully a U-turn is coming.

The better question: What are Mississippi and other southern states doing right?

Quoting the Boston Globe:

“Meanwhile, leaders first in Mississippi and then in neighboring states embraced — and enforced — a phonics-driven reading curriculum and a back-to-basics math approach. State leaders across parties and administrations demanded that all students must be learning, with a threat of consequences for districts and schools that failed.

Mississippi plowed ahead with controversial reforms, including a reading test students must pass to proceed to fourth grade. Those efforts have borne fruit, producing first the “Mississippi miracle” and in more recent years what’s been dubbed the “Southern Surge,” also encompassing Louisiana and Alabama.”

In short, these states embraced what they call “evidence-based learning.” That includes phonics-based reading education and targeted help for struggling kids.

They’ve also—critically—set and enforced high standards, only letting kids move up grades once they’ve mastered the material.

The Mississippi Miracle is huge. Not only does it show we can vastly improve schools, it shows we can vastly improve the worst schools—those that were dealt a challenging socioeconomic hand—without spending a fortune.

But as rational optimists, we’re not satisfied with just getting back to 2013 levels of achievement.

School can be better than it’s ever been, for every kid, everywhere.

We’ve long known how to give all kids a world-class education.

In 1984 University of Chicago professor Benjamin Bloom confirmed what aristocrats knew for centuries: Kids learn best with a dedicated, one-on-one tutor.

Bloom found just seven weeks of personalized tutoring could transform C students into straight-A students—from average to exceptional. Critically, almost any student had the potential to achieve top grades.

But few parents can afford personal tutors. So one-size-fits-most schools were the best we could do. Until now.

What happens when the cost of a personal tutor falls to almost zero? We’re on our way to finding out.

I visited and met with the founders and students of the premier “AI-first” school, Alpha School in Austin, TX. “Impressive” doesn’t begin to describe it.

Alpha kids learn 2.6X faster than the average student and achieve in the top 1-2% in every subject. 96% of Alpha students say they love school. Half even prefer school to vacation!

Alpha uses AI to address the single biggest weakness of traditional schooling: personalization around a student’s current ability. Forget futuristic AI tutors for a moment. Simply tailoring every student’s schoolwork to the right level of difficulty improves outcomes by leaps and bounds.

In a regular classroom of 25 kids, the teacher must teach to the average. Kids who find it too hard get frustrated and disengage. Kids who find it too easy get bored and disengage.

The solution is tight personalization around a student’s current ability. Alpha School President Joe Liemandt emphasized to me how critical this is. AI is perfectly suited to deliver it.

Alpha’s suite of AI tools constantly adjusts to the learner’s speed, keeping lessons within 10% of the student’s “zone of proximal development.” This ensures students struggle, but productively. Productive struggle is the sweet spot that unlocks learning.

One case study followed seven boys who’d fallen two years behind in public school. At Alpha, these boys caught up two entire grade levels in just six months.

But many are skeptical of AI learning. Having followed its development for years, I hear this pushback a lot:

Really, kids on laptops all day?

Didn’t we learn our lesson during COVID?

Alpha has pioneered “2-hour learning.” Every morning, kids do 25 minutes each math/reading/science/writing, with a short recess break between each. The goal is to finish academics in two hours, then do something fun and meaningful the rest of the day.

Afternoons at Alpha are spent on “life skills.” Fifth graders ran a food truck and made $4,000 in profit. Another class managed a real Airbnb. Some afternoons are robotics labs; others are chess tournaments.

Stephen McBride, Matt Ridley and I visited Alpha’s flagship school in downtown Austin on a normal school morning. We walked past a group of 10-year-old boys sitting together on a big blue couch, chatting as they worked on their laptops.

Then we talked to three high school seniors. One had just become the first teenager ever to produce a live Broadway play. Another built a custom chatbot that helps teenage boys who struggle talking to girls. The third was accepted to Stanford.

All three beamed with passion and charisma I wish I had at age 40, let alone 17.

I want this for my kids!

Sitting with Joe Liemandt, he emphasized:

The AI designing personalized lesson plans is huge. But it’s not the most important thing. The real secret sauce is the motivation and incentives.

Yet would you believe…

Paying kids for good grades is less popular than spanking.

77% of Americans oppose paying kids for good grades—while 74% oppose spanking.

That’s a shame because paying kids works—if you structure it right. Harvard professor Roland Fryer studied this comprehensively. In one experiment, he paid kids $2 per book read and boosted their reading by 40%. In another, kids paid to attend class were 22% more likely to show up.

But there’s a catch. Fryer found that you must pay for specific, controllable inputs. Read one book, get $1. This works. But if you pay kids $5 to achieve an A+ grade, that tends to fail. Rewards should be for inputs students can directly control.

It’s these little details that matter, which is why I’m so optimistic that schools like Alpha are experimenting and innovating with motivational frameworks.

Alpha incentivizes learning in several ways. Kids can unlock gift cards for completing hard lessons. They can earn “Alpha bucks” to buy event tickets, toys, or start an investment account.

But it turns out the best motivator isn’t monetary. Alpha uses a program called TimeBack that incentivizes kids to “crush” their core academics in two hours a day. They can then choose a project that interests them. Some challenging projects that 2^nd graders completed:

• Build a drone and land it on your head
  

• Beat your parents in a spelling bee
  

• Run a 5k in less than 35 minutes
  

• Present the news like an anchor in front of the whole school

Alpha middle schoolers took a trip to Poland to help educate Ukrainian refugee kids. As you can imagine, these kids had been through a lot, and getting them to engage with school was tough.

What worked best? A “learn and earn” program that earned them $2.50/day for completed lessons.

Alpha School isn’t meant to be perfect. It’s an innovation lab. Last year it began offering a version of its AI software bundle called “Alpha Anywhere” for families to use at home. It’s the exact same software Alpha students use on campus.

It didn’t work. Students who used it showed no improvement over normal school. The AI program alone lacked the incentives, environment, and most importantly, the human element that really make Alpha School special.

Alpha Anywhere has been tweaked and improved since then, and it’s generating better results. I’m skeptical it will ever hold a candle to the best in-person education. But it doesn’t necessarily have to make a huge positive impact.

AI learning elevates human teachers

In Neal Stephenson’s sci-fi novel The Diamond Age, a girl named Nell is born into poverty that’s supposed to be inescapable. But her brother steals an AI-powered educational device for her. It transforms her life by teaching her through personalized, interactive stories.

Spoiler alert: While the device has almost-magical AI, there’s also a real human on the other side. She performs voices, mentors Nell, and eventually forms a strong bond with her.

Top-tier education will be the same in real life: It takes AI plus great teachers.

“What about the teachers?” is a common question when discussing AI schools. Teaching is a HARD job. Not only must they try to teach 25 unique kids with a one-size-fits-all lesson. Teachers are bogged down grading papers, designing lesson plans and managing behavioral problems.

AI can take care of all that, freeing teachers to do what matters most: inspire, mentor and motivate students.

At Alpha, teachers are called “guides.” They don’t lecture or grade papers. Instead they mentor students, helping them build confidence, overcome problems, and encourage curiosity. Alpha guides often come from fields like coaching, entrepreneurship, or global education. They’re paid well—a minimum of $100k.

At this point you’re probably wondering what Alpha School costs. Its Austin campus is $40,000/year. Its new Virginia campus is $65,000/year.

So yes, there’s strong selection bias at play. Alpha students tend to come from wealthy families. For now, Alpha primarily elevates the “ceiling” of education by helping fortunate kids reach their maximum potential.

But there’s every reason to believe AI schooling will play out like most technologies. Competition will force quality up and prices down, raising the “floor” of education and giving access to all. We’re already seeing glimmers of this.

The World Bank studied 422 high school students who used ChatGPT to boost their English grammar, vocabulary and writing skills over a dozen 90-minute sessions. The students made learning gains equivalent to two years of traditional schooling—in six weeks! Girls and struggling students saw the largest gains. AI schooling can be a great equalizer.

The human element of great teachers/guides will always be hard to scale. But software is nothing if not scalable. Within a few years, why couldn’t every family have access to the same quality AI-schooling tools Alpha Kids have?

Alpha is also working to expand access. They opened a full campus in Brownsville, TX, one of the US’s poorest towns along the US-Mexico border. Tuition is only $10,000, and most families receive assistance. They were able to open a school there because SpaceX’s Starbase, half an hour away, brought an influx of SpaceX employees to the area.

Results are strong. The average Brownsville student learns 2.1X faster than at a standard school. An all-girl team from Brownsville beat the Austin campus in a self-driving car coding competition.

Not everyone agrees though. Wired wrote a hit piece on Alpha Brownsville, featuring parents who felt the school’s standards were too high and uncompromising.

There’s no question Alpha has high standards. Liemandt is unapologetic about it, saying: “The key to your kids’ happiness is high standards.”

This theme of high standards comes up again and again—not just at Alpha School, but across nearly all schools that produce great results. Alpha, Mississippi, and the other high-performing school systems you’ll meet momentarily vary a lot. But the one thing they all have in common is they embrace high standards.

Now let’s meet some other categories of AI-forward education innovators. What ties these together is they all…

Push the boundaries of what makes a great school.

Innovative Public Charter School Networks

The schools in this category are the closest thing to Alpha. They perform highly and are integrating AI-driven personalization more and more into the curriculum.

There’s one huge difference from Alpha: They are tuition-free, and admission is a lottery.

To varying degrees, these schools let AI do a bit of the academics while teachers lead small groups or provide one-on-one help and mentorship.

Prominent schools in this category:

• Summit Schools (California and Washington)

• Rocketship Public Schools (California, Tennessee, Wisconsin, Washington, D.C.)

• Navigator Schools (California)

Notice they all operate in California, which has great regulatory flexibility around charter schools that permit bold innovation.

“Charter school” is a bad word in some circles because not all kids can attend. In fact, many people dislike any option besides public school. In my opinion, anything that encourages innovation and competition in schools is a great thing.

Innovative Private Schools

Khan Lab School is the gold standard here. Sal Khan, whose Khan Academy YouTube lessons have helped millions of self-learners, is the founder.

Khan Lab’s explicit purpose is to push forward innovation in K-12 education. It calls itself a “living laboratory” and an “incubator.” The school’s core goal is to invent, refine and share educational innovations that can be adopted by other schools.

It’s safe to say Khan Lab has experimented more than any other school since its founding in 2014. So it’s telling that Khan emphasizes two concepts above all others: “The personalization of pace with high expectations.”

Khan Lab students use AI extensively through the Khanmigo app (which Alpha School also uses as a core tool). Teachers provide not only mentorship and guidance but subject matter expertise. Khan Lab hires teachers with deep expertise in art, math, history and so on.

The academic results are fantastic. Kids consistently score in the 95^th-99^th percentiles on standardized tests. But like Alpha School, there is strong selection bias. Khan Lab costs $30,000+ and its two campuses are in two of the wealthiest places on Earth—Mountain View and Palo Alto.

California is also home to many exceptionally innovative private schools that aren’t tech-focused. For example Red Bridge in San Francisco emphasizes developing students’ agency. Teachers serve as guides and mentors. And instead of grouping students by age into grades, students must advocate for their own “promotion” to the next level.

There are many thousands of innovative private schools around the world. They’ve all got glossy marketing brochures that sound amazing, and I’m sure many of them are. But it’s hard to judge their models from afar due to selection bias. Alpha and Khan Lab stand out as being highly public with their results and sharing how they continually tweak their models to get better.

Microschools

These are tiny schools, often 20 kids or fewer, that tend to be a hybrid of homeschooling with some traditional structure.

There are 95,000 microschools in the US. Three notable tech-forward ones:

• Primer is building a network of small, in-person learning centers run by passionate teachers who think more like entrepreneurs.

At Primer, the morning is dedicated to core academics. Students use laptops to move through personalized math and reading tracks at their own pace. The AI adapts to each child’s level in real time.

The afternoon is for “Pursuits,” where students dive into passion projects (sound familiar?) like starting a Shopify store for slime or creating a YouTube science series. Students are encouraged to use AI to generate ideas, branding, and business plans.

Primer is what happens when you build school around a kid’s individual curiosity. It’s possible to start your own microschool.

• Acton Academy aims to turn a student’s boring school experience into a “hero’s journey” where guides nudge students toward finding solutions on their own. Similar to Alpha, Acton crams a full curriculum into two hours of efficient learning. Then the heroes are set off on afternoon projects like learning musical instruments or building startups.

• Prenda lets kids meet and lead their own class, with the help of a local adult guide. Anyone can order the materials from Prenda to start their own Microschool.

Cyberschools

This category of schools is purely virtual.

Unbound Academy is notable here as it was founded by Alpha School co-founder Mackenzie Price. It uses the same 2-hour learning framework, but in a virtual setting. The other four hours of the day will be for mentorship, student projects and workshops on topics like public speaking and financial literacy.

Unbound just started its first school year this fall, in Arizona, which accepted its application after many other states rejected it. But it’s not the only cyberschool offering kids (and parents) a hybrid “2-hour school day.” Startups like Novatio are following the same playbook of short, personalized AI lessons.

I’m skeptical of purely virtual schools, especially after our brief trial run during COVID lockdowns. I would be happy to be proven wrong.

Parents’ corner: Do-it-yourself education enhancers

You don’t have to change schools to have a positive impact on your kids’ education.

We’re lucky. None of our ancestors had the opportunity to take advantage of AI tools to enhance education. We do, so let’s use it.

There are dozens of options, and more are launching all the time. I’ll only comment on the ones I’ve personally tried.

Ello

Last year my son Jack (7) was doing well in 1^st grade. His teacher assured us he was in the highest reading group all year. Then we got his test scores. Math was good, reading was bad.

To help him catch up, I tried out four different AI reading programs and settled on Ello. Jack reads aloud with it for 20 minutes every morning on an iPad. It listens to him, gently corrects his mistakes, and only moves ahead once he’s mastered a concept.

I can see his progress through a parent dashboard. It helped him jump ahead 1.5 grade levels in reading in three months. It only costs $15/month.

Do I have a tinge of dad guilt that I should be reading with him instead of an iPad? Yes, and I read with him often, too. But now he has three “jobs” he fully owns every morning: getting dressed, brushing his teeth, and reading two books with Ello, which takes about 15 minutes. It’s built his confidence and helped teach him independence and responsibility.

Ello gets an A+ in the Steinhart household.

Lexia

Lexia helps kids learn reading and language at their own pace, using a rudimentary form of AI to tailor the difficulty level to each kid.

Jack uses Lexia at school, and he loves it so much we installed it on a home iPad too. It’s gamified, and he competes with his classmates to reach the highest levels. Then his teacher can see what skills each student has worked on (or needs more help with) from a personalized data dashboard.

Highly recommend.

Synthesis

An “interactive tutor,” born out of the experimental school Elon Musk created for his kids at SpaceX.

Its pitch is teaching students how to think. Instead of drilling kids with facts or multiple-choice questions, Synthesis throws them into fast-paced simulations with no easy answers.

I tried this with Jack and was disappointed. I believe there’s not enough structure, and kids have too many options. Jack kept playing the same math blaster game over and over again, then lost interest.

But remember, this is the worst these tools will ever be. AI is only getting better.

Old-fashioned paper books

My 4-year-old son Max is just starting to read. He’s not ready for the iPad yet. The most effective tool I’ve found, by far, is Julia Donaldson’s Songbird collection of books.

Each book is very short and only introduces a handful of new words, building the new reader’s confidence. The series is organized into levels 1 through 6. On level 1, he’s gone from zero to being able to read about 20 words and some short whole books. Most importantly it’s built his confidence and is cultivating an enjoyment of reading.

Highly recommend.

Learn Your Way

This is Google’s augmented textbook. It transforms boring texts into a personalized, interactive learning experience. You can upload a lesson about anything, and this “living textbook” personalizes it to be engaging and effective for your kid.

If your daughter loves basketball, she can learn long division from an interactive LeBron James.

I have not tried the full version yet (I’m on the waitlist). But the demo videos look extremely promising. You can look at a few prepackaged examples here, like a lesson on economic systems tailored for a soccer-loving kid.

Teacher’s assistant apps

Another useful category to keep an eye on. The fewer repetitive, mundane tasks teachers have to do, the more personalized coaching they can give.

Khanmigo is Khan Academy’s AI-powered tutor, built in partnership with OpenAI. It’s free for teachers, courtesy of Microsoft. Without it, teachers spend 40%–60% of their time on admin and prep. It acts as a tireless teaching assistant, capable of drafting lesson plans in minutes and providing personalized feedback on student work.

Parents can also get Khanmigo for $4 a month to help supplement school, or help manage a homeschool curriculum.

This is just the beginning

AI-enhanced education is in its infancy. It’s only going to improve, fast.

If you’re a parent or grandparent, above all else, try stuff. Demo the latest AI tools. Talk to ChatGPT. See which AI tools resonate with your kids. It helps to live in a place that allows and prioritizes innovation in schools, if you can.

Eventually, the benefits of AI-enhanced education will percolate to all. But for the next 5 or so years at least, it’s up to us as parents to make sure our kids benefit.

And keep in mind, the greatest benefit of AI-enhanced education is personalization. Kids learn differently, and different kids like to learn about different things.

AI-enhanced education can help find your kid’s unique spark of curiosity, ignite it, and keep that flame growing.

We only get one chance to educate our kids. Let’s do it right!

—Dan Steinhart