On Joe Liemandt on Alpha School

Joe Liemandt, the funder, principal, and (along with Mackenzie Price) driving force behind Alpha School, recently gave a lengthy interview to Patrick O’Shaughnessy that went into a fair amount of detail about the thinking and the technology behind Alpha School and related ventures. It joins a growing list of material about Alpha School (see my “Further reading and listening” section below). I want to focus on this interview in particular, since as the chief funder Liemandt is ultimately responsible for the direction that Alpha School takes, and thus will bear a large part of the responsibility for its overall success or failure. I’ve supplemented it a bit with Alpha School-related material from other sources that provides additional context for Liemandt’s remarks.

A disclaimer: I am not an expert on K-12 education. I am an adult learner who has used and reviewed online learning systems from Math Academy and Physicsgraph and liked them. My goal in this post, as in my others, is to be neither an uncritical cheerleader nor a dismissive critic, but rather to provide a measured look at the topic and to try to add some value to the public discussion of it.

You can watch the video for yourself or read the transcript, so I don’t intend to recap the interview or even discuss every part of it. The following are the areas I found most interesting.

Initial impressions

The video is prefaced with a clip from later in the interview:

All educational content is obsolete. Every textbook, every lesson plan, every test, all of it is obsolete because GenAI is going to be able to deliver a personalized lesson just for you.

These are Joe Liemandt’s words, but it was Patrick O’Shaughnessy who chose to highlight them. In doing this I think he gave the audience a misleading impression regarding Liemandt’s ideas and what Alpha School actually is. Taken at face value and in isolation, this is a standard pitch we’ve heard a hundred times before: “Everything you know about education is wrong, ChatGPT can teach everything your child needs to know.” The real story here is more interesting; in particular, Alpha School is using LLMs, but not in a conventional way, i.e., as chatbots.

After listening to the Liemandt interview, I listened to a podcast interview with Mackenzie Price. Consistent with someone who hasn’t done public interviews for many years, Liemandt’s comments are rather unpolished compared to Price’s, which come off to me like a very practiced sales pitch. I have nothing against sales pitches (I’ve worked in enterprise software sales groups for many years), but here I give the edge to Liemandt in terms of being perceived as an authentic and credible spokesperson for Alpha School.

“Time back” vs. ”2 hour learning”

Early in the interview Liemandt makes the point that one major way to make children like school more is just to have less of it:

My first week as principal 3 years ago, you know, I’m now accountable to the kids . . . and so I’m talking to some fifth graders I’m like . . . “What would make you love school?” and they’re like, “less school,” and I’m like, ”Okay, well how much less?” and they’re like, “no school,” and I’m like “Okay, well, that seems a little light as a principal to like have no school,” and so basically we negotiated . . .

He then expands on the central pitch of Alpha School: that students need to spend only 2 hours a day doing hard-core academic learning, and then can spend the rest of the day in more enjoyable pursuits. Later in the interview he puts it this way: “Give kids their time back, . . . that is by far the number one motivator for kids.”

Alpha School emphasizes the “2 hour learning” pitch (even to the point of having an associated website 2hourlearning.com and company 2HR Learning, Inc.), but my personal opinion is that “giving kids time back” is a more effective and defensible marketing message. Like Benjamin Bloom’s “2 sigma solution,” “2 hour learning” is a slogan that’s superficially attractive but typically ends up being qualified in one way or another, often by people disinclined to believe the hype: “It wasn’t really 2 sigma, because there are factors A, B, and C that made Bloom’s study unrealistic.” “It’s not really 2 hours a day, it’s really more like 3.5 hours a day, because it doesn’t account for X, Y, and Z.”

Why provide an opportunity for people to nitpick your claims and muddy your message? From a kid’s point of view, shortening the traditional school day even a bit is a win—recall how happy you were as a child when school ended an hour early.

TimeBack, Incept, and “AI Powered”

Perhaps the most interesting section of the interview is Liemandt’s discussion of TimeBack and Incept, two of the software products being developed for Alpha School and related ventures. Together with comments by the pseudonymous Austin Scholar and @turing_hamster (see below), these give a more detailed picture of how the academic component of Alpha School works.

The first, and most important, point is that Alpha School does not use LLM chatbots like ChatGPT and its ilk. In fact, Liemandt in this interview and Mackenzie Price in hers go out of their way to emphasize that LLM chatbots as they currently exist are not a suitable base on which to build a learning system, because students will primarily use them to cheat. To quote Liemandt:

[Deploying ChatGPT is] literally the worst thing you could do. Just to be very clear, you are not a principal at a school if you think ChatGPT is good. 90% of the use of ChatGPT is to cheat. . . . Now, there are 10% use cases of people who are like, “I need a Socratic tutor where I’m super excited about it and I need to dive in deeper,” right? And that 5 to 10%, those kids are going to crush it with it, and you should give it to those 5 or 10%. 90% of the kids are like, “How do I use this to cheat?”

To revisit one of my pet peeves: This is all well and good, but when you advertise Alpha School as an “AI Powered Private School” (the tab heading on its website), talk about your “AI tutor” (on the 2hourlearning.com website), and have press articles with titles like “A.I.-Driven Education: Founded in Texas and Coming to a School Near You,” people can be forgiven for thinking that you’re just another undifferentiated entry in the increasingly long line of startups trying to take ChatGPT (the one ”AI” product the typical person knows about) and teach kids with it.

I thought over-emphasizing the “AI” aspect in marketing was a mistake when Math Academy did it, and I think it’s a suboptimal approach here as well, especially as we’re seeing more stories about the downsides of chatbots: “No, no, no, our AI is not like their AI; ours is the good kind of AI.”¹ Now back to Liemandt . . .

We literally don’t have chat functionality activated in our AI. Like it is not a chatbot. It’s terrible . . . that’s not what people need. . . . Where we use the AI is generating personalized lesson plans and personalized lessons themselves. . . . That is a great use of AI. A second great use of AI is actually to watch the screen and coach the kids, right? To give them the game film.

As I understand it, Incept is the component within the overall software stack that generates the lessons themselves. An LLM is used in that context, not to interact directly with students but to generate “variations on a theme,” that is, sets of questions and answers formulated from underlying templates but customized in various ways. (For more detail, see this tweet; as noted in the ensuing thread, the PhysicsGraph team is doing something similar.)

TimeBack is the overall software stack (or “EducationOS” as the thus-far-minimal TimeBack website terms it) that integrates the various learning apps, both those produced in-house (e.g., AlphaRead) and those sourced from others (e.g., Math Academy). By analogy to Math Academy and PhysicsGraph (the systems I’m most familar with), this presumably includes taking students through initial diagnostic exams, maintaining an overall knowledge graph and students’ progress against it, determining course and lesson sequencing for each student, and providing status dashboards for students, parents, and “guides” (Alpha School’s version of teachers).

Monitoring students

As noted by Liemandt, TimeBack also incorporates an LLM-based system that tracks students’ actions on video and attempts to identify times when students are not being attentive to lessons or otherwise not completing their work. Liemandt explains:

So 3 years ago, we would just sit there and have teams of people all night, you know, the kids would finish and then all night you would just sit and watch the videos and annotate them of, okay, that’s what’s going wrong here, this is what the kids [are] doing. . . . It’s just building out a video model, a reinforcement learning video model, . . . with human feedback, and the human feedback is not from the kids, it’s from people watching it to say, “I need to train this up on what how kids learn and what is effective learning, what are the anti-patterns, and what do we want to coach for.”

Between this and other uses, per Liemandt, Alpha School is apparently spending on the order of $10,000 US per student per month on LLM usage.

This sort of fine-grained student monitoring is exactly what edtech critics like Audrey Watters and others decry as “techno-fascism,” “surveillance capitalism,” and the like—the idea that “tech bros” want to turn every school into a panopticon in which a student’s every move is monitored and measured. (And, presumably, students who chafe under the yoke are removed from the classroom, in the first stage of the “school-to-prison pipeline”.)

Yet, as Liemandt’s mention of “game film” suggests, such close monitoring of students, including automated measurement of what they do, is not uncommon at all in other school activities, most notably school sports. What makes this different? Perhaps at least three factors:

First, the monitoring is compulsory. Students decide for themselves whether they try out for a school team, but they have no choice as to whether or not they go to school, and little or no choice regarding the regime they’re under once they get there. Any choice being exercised is by their parents, who decided to enroll them in Alpha School in the first place.²

Second, there are concerns about misuse or commercial use of data generated from student monitoring—that somewhere buried in a 50-page terms of service agreement is a clause granting Alpha School or its affiliates blanket permissions to sell any and all student data to all comers.

Finally, some might worry about collected student data being used to justify expulsions or other lesser punishments—for example, by an Alpha School-affiliated charter school trying to get rid of unwanted students to improve its overall test scores. Being kicked out of school has a lot greater consequences than being kicked off the football team or the marching band.

How Alpha School responds to these concerns is up to it. There are some people who will never be convinced of its beneficence in this respect, but I think it’s worth reassuring its most important audience, parents of prospective students.

Money as motivator

Liemandt lays great emphasis on student motivation as the most important factor for Alpha School to focus on:

Motivation is 90% of the solution. Okay. The edtech is 10%. . . . We keep putting edtech into schools and test scores keep going down. . . . The issue is edtech is necessary but not sufficient. . . . Educators will say you need two things to teach a kid. First, and most important, a motivated student and then, second, you need to put them in lessons of the correct difficulty. . . . Edtech does the second one really well. . . . But are you going to engage? Do you have any interest in engaging? Right? Or if I don’t engage with the app, right? Then I’m not going to learn anything. Motivation is 90%. And Alpha is 100% built around solving the motivation problem.

Having to spend less time on academic subjects is a powerful motivator, as discussed by Liemandt elsewhere in the interview. Alpha School also employs another motivational technique, rewarding students who perform desired tasks with “Alpha bucks,” a private currency that students can exchange for various treats.

In the interview Liemandt goes further, discussing how he rewarded students with relatively large cash prizes for acing Texas student assessment tests:

Back to money motivation: I’m like, “I’ll give you a $100 bill for a hundred on the Texas STAAR, 100 for 100.” They’re like, “That’s still impossible, I can’t get one.” I’m like, “No, no, no, there’s a catch: any grade level.” . . . And they’re like, “Okay, I’ll take a third grade test.” Yeah. And I’m like, “Great.” And so they go take a third grade test with STAAR, they get $100.

Liemandt continues the anecdote, explaining how the students then repeated the feat by taking the fourth-grade STAAR test, still below their grade level, scoring 100 again and winning another $100 US.

This is interesting because, as referenced in the AstralCodexTen review of Alpha School, research by Bradley Allan and Roland Fryer indicates that rewarding students for “outputs” (in this case, perfect test results) is less effective than rewarding them for “inputs” (e.g., reading a book). From that point of view, Liemandt’s approach is less than optimal.

But let’s continue with Liemandt’s anecdote:

They take [the fifth-grade test] and they get an 85, right? 75 or 85. And I’m like, . . . “Do you want the AI tutor to generate the lessons so you can get 100?” And they’re like, ”Let me see how many.” And then they see it and they’re like, “I can do that in a week.” And they’re like, “I’m in.” And then they get a hundred, right?

Back to Allan and Fryer:

In our experiments, input incentives were more effective than output incentives, suggesting that students do not know how to increase their test scores. If students only have a vague idea of how to increase their test scores, then when provided with incentives for performance, they may not be motivated to increase effort.

The implication of Liemandt’s remark is that incentives for outputs can indeed motivate students, if they are given specific instructions on exactly what they need to do to achieve those outputs, for example, by going back and mastering prerequisite topics that the system has identified that they lack.

Trilogy, ESW, and funding Alpha School

In the latter part of the interview, O’Shaughnessy prompts Liemandt to go down memory lane and talk about his experiences at his companies Trilogy and ESW Capital.³ I was working for enterprise IT companies in the 1990s on and had previously never heard of Trilogy, much less ESW Capital, so this section held particular interest for me. ESW Capital, apparently a spin-off from Trilogy, is a private equity firm investing in software companies,

The popular image of private equity firms is as “vulture capitalists” that buy companies, load them up with debt in order to pay themselves dividends, lay off most of the staff and strip the companies of assets to further enrich themselves, and then stand back as the companies slide into bankruptcy and lay off the rest of their employees.

From my perspective, ESW Capital seems more like a modern incarnation of Computer Associates, once the second largest software company in the world. The basic model is to acquire companies that are no longer growing strongly but have substantial installed bases, and then to switch the companies’ approaches from “hunting” to “farming.” That is, the new focus is on supporting existing customers in order to realize revenue from subscription renewals and expansion deals, and cutting back on those areas of the business that don’t directly support that strategy.⁴

When done well this can be a very lucrative strategy, one that throws off a large ongoing stream of cash. As Liemandt remarks:

What you’re just doing is managing, and . . . you’re pulling the cash out to do it and there is investment you make because that extends it. Software still is one [area] where if you focus on the customer—I mean, just basics, right—that you can extend . . . the area under the curve [i.e., total revenue per customer]. . . . And then you get some nice things like, oh, an AI uplift and, you know, things like that that that allow you to do it.

What does this have to do with Alpha School? First, it provides money (lots of money) to create not just Alpha School but the whole panoply of software and related technology for which Alpha School is the first customer, as well as to expand and extend the Alpha School model in various ways:

I was like, “You’ve been telling me for 25 years that you could run this company better than me. Let’s see. Right. Good luck. I’m going to go be a principal.” . . . And I said, “And by the way, I’m taking a billion dollars so . . . figure out and get that going.” And so now I turned from, you know, boss to shareholder. And literally I’m like the worst shareholder now in the world. . . . I’m like, “Guys, I need all this capital. Like I need more cash. . . . You know this expansion that Mackenzie [Price] wants to do, right? It’s going to take $300 million.”

Alpha School also benefits from a training model originally pioneered by Trilogy, in the form of Trilogy University, and now embodied in Gauntlet AI.⁵ The Gauntlet AI program is essentially an internship (unpaid except for room and board) in which participants work 80-to-100-hour weeks for 10 weeks, with the promise of guaranteed employment for those who graduate (about half of those who start). Presumably a good number of those graduates end up in the Joe Liemandt keiretsu, of which 2HR Learning, Inc., Alpha Schools, LLC, TimeBack, LLC, and various other Alpha School-related ventures are members.⁶ (Alpha School also recruits through Crossover, apparently another member of the keiretsu.)

Growth-maxing and business models

I get the idea that underlying Joe Liemandt’s interest in Alpha School and related ventures is not just a desire to transform education for the better (a desire that strikes me as entirely sincere), but also a desire to go back to his Trilogy roots and build a business from scratch. Certainly his ambitions in this area are outsized: not just to run a few private schools but to stake out a major position in what he refers to in the interview as a trillion dollar market.

(Is this a bad thing in principle? Some people are very allergic to the idea of for-profit companies being involved in education, for example, pointing out that for-profit colleges encouraged students to take on excessively-large and non-dischargeable student loans for degrees of dubious worth. However, that was arguably in large part a consequence of misaligned government incentives and regulations, which many nonprofit institutions also took advantage of to grow enrollments. More generally, I don’t think nonprofit status confers any special virtue on educational institutions; each institution, nonprofit or for-profit, should be judged on its own merits.)

The overall strategy pursued by Liemandt, Price, et al., seems to be to offer a set of differentiated education products, which we may think of as the Gold, Silver, and Bronze versions of Alpha School:⁷

First is Alpha School itself, both the original in Austin, Texas, and its sister schools being established across the US. This offering, at a price point ranging from $40,000-75,000 US per student per year, is intended for affluent families with parents who are relatively tech-savvy (so as to appreciate the “AI Powered” messaging) and willing to consider nontraditional schooling.

Based on the Alpha School location list, there are 15 of these schools existing or planned across the US (along with the lower-cost Brownsville, Texas, campus, which is something of a one-off). Some of these are for lower grades only, perhaps due to a combination of limited staff and the desire to have students grow within the system. There are also variants of Alpha School targeted at different student profiles and offered under different brands: GT School for more academically minded students, Texas Sports Academy for student athletes, Next Gen Academy for gamers, and so on. To the best of my knowledge these have comparable pricing to the original Alpha School.

The Alpha Schools and their private school variants will likely be a “halo product” for the entire 2 Hour Learning project: The selective effects of the relatively high tuition will likely mean that these schools will produce the best academic results, they will feature the most impressive (and expensive) afternoon projects and extracurricular perks (e.g., overseas trips), and are likely to be the schools most featured in student testimonials, pundit newsletters and tweets, and national press articles.

Next come a potpourri of “Alpha School lite” offerings, apparently priced in the $5,000-15,000 US range. These include virtual charter schools (like Arizona’s Unbound Academy), systems for homeschoolers (Alpha Anywhere), and programs to supplement traditional schooling (Prequel). Except for Prequel, these aim to leverage government funding to bring the cost down to a level more affordable for less affluent families, either via direct funding of charter schools or indirect funding through education voucher programs. For example, the Texas Education Savings Account Program will provide almost $11,000 US per student per year to offset the cost of private schools, and up to $2,000 US per student per year for homeschooled students.

Liemandt also talks about plans to expand this model to more schools, whether directly affiliated with Alpha School or just licensing the TimeBack platform (e.g., through the 2 Hour Learning for Private Schools program):

I’ll give you an example of one of one of the ones [that’s] not funded yet, but the business plan is super awesome, which is, Texas just passed school choice, [a] billion dollars [in funding]. . . . This group is . . . building Texas Sports Academy. Their goal is to go to every school district in Texas . . . and build a . . . micro middle school for D1 athletes. . . . They’re going to be based on the . . . Timeback platform. And then in the afternoon, it’s going to be one of the local coaches, right, that everybody in town reveres, who is going to have a . . . middle school . . . for people who want to be D1 athletes, and they’re going to spend the afternoon doing whatever awesome athletics. . . . As we prove this out in scale there’s actually a lot of . . . famous professional athletes who are like, “I would love to start a school like that.”

The final offering is the most ambitious: to reduce the cost of TimeBack, Incept, and the other components of the Alpha School software stack to the point where it can be offered to most children around the world:

I was talking to some ministers of education of other countries and, you know, I was like, “Look, you know, sometime in the next 5 years—I don’t know exactly when—but there’s going to be a tablet that’s less than $1,000 that is going to teach every kid on this planet everything they need to know in two hours a day. . . . Our AI burn is like 10,000 bucks a kid now, but it’s going to be down. It’s going to be for everybody.

Now, it’s not as if we haven’t been down this road before, although inflation has raised the price from $100 US to $1,000 US; we’ve seen the TED Talk and read the post mortems. But even if the technical and cost obstacles can be overcome (and I’m not going to bet against it⁸), we go back to one of Liemandt’s first comments: “The edtech is 10%. . . . Motivation is 90%.” So the “$1,000 tablet” by itself will not be sufficient.

From guides to games

In the original Alpha School model, motivation is provided by highly-paid guides. What’s Liemandt’s answer for motivating a billion children? The answer appears to be video games:

The people who know how to engage kids more than anything is video game developers. . . . When I talk about this tablet, I absolutely believe that there will be video games on that tablet that are totally engaging to kids that they will go through and it will teach them everything they need to know. . . . We’ve invested over $100 million in getting some of the best video game designers in the world. They’re down in Austin, and they are building a video game on top of the 2-hour learning engine that they believe will rival any video game on the planet. And the difference [is], parents will be happy to let their kids play it and get their kids to [the] top 10%.

Again, it’s not as if Alpha School and 2HR Learning are the first organizations to think of gamifying education. We’ll have to wait for wider release of the full TimeBack stack to see how their approach compares to that of others.

Is there another way?

As related by Liemandt elsewhere in the interview, Trilogy was originally a scrappy upstart, just Liemandt and his co-workers in the proverbial Silicon Valley garage. The contrast with Liemandt’s present day project could not be greater: The overall Alpha School/2HR Learning/TimeBack project has hundreds of millions of dollars in funding and hundreds of developers. It’s taking on the most difficult challenge in education, transforming K-12 schools, and finding the task far from simple. (For example, the affiliated venture Unbound Academy has had almost all of its applications to run charter schools rejected.)

The two systems I’m most familiar with, Math Academy and PhysicsGraph, are taking a very different approach. They’re built by small teams (3 people for PhysicsGraph, perhaps a dozen or 2 for Math Academy), appear to be primarily self-funded, focus on defined topic areas, are one or two orders of magnitude less expensive ($200-500 US per year per student), and are targeting much more tractable markets (supplemental instruction for K-12 and postsecondary students, and career- and hobby-related training for adult learners).

The result is systems that embody the same general principles of academic learning as Alpha School (as laid out in Justin Skycak’s book The Math Academy Way) and are (in my experience) very effective. Math Academy in particular is so well thought of that Alpha School itself adopted it for use within the overall TimeBack system. (Although it’s always possible that it will be replaced later with a system developed in-house.)

Which approach will be more successful, Alpha School’s full frontal assault or the more guerilla approach of Math Academy, PhysicsGraph, and their imitators? I’m afraid that I’ll have to end on an ambiguous note and say that I just don’t know, although I’m looking forward to seeing both approaches play out. In the meantime I’ll keep checking up on Alpha School and its various offshoots, even though I don’t have a school-age child to consider enrolling.

Further reading and listening

I’ve divided the following links into three sections, as follows:

Alpha School tells its own story:

• Joe Liemandt interview with Patrick O’Shaughnessy for the “Invest Like the Best” podcast. The subject of this post.
• The Future of Education. Mackenzie Price’s newsletter.
• “Class Dismissed” by Jeremy Stern for Colossus magazine. A profile of Joe Liemandt and Mackenzie Price, written with their cooperation. (Colossus is part of the media franchise that includes “Invest Like the Best.”)
• “A.I. School Is in Session: Two Takes on the Future of Education” with Kevin Roose and Casey Newton for the New York Times. An dual interview with MacKenzie Price and Princeton professor (and historian of science) D. Graham Burnett.
• “Khanversation #53: Alpha School and and education in 2025” with Razib Khan and Josiah Neely. An interview with Pamela Hobart, who began as an Alpha School parent and now works for Alpha School.
• @turing_hamster on X. The social media feed of a developer working on the software used by Alpha School. Although it’s scattered across multiple tweets, this is probably the best place I’ve found so far to get more insight into the technology behind the TimeBack system.
• “Welcome to the Future of Education” from 2HR Learning. A white paper describing the “2 hour learning” system. It has lots of material covering claimed results, and a bit about the actual apps.
• “2024 Cyber Charter School Application for Unbound Academic Institute.” Unbound Academy’s application to open up a virtual charter school in Pennsylvania. (The actual application begins on the page headed “Cyber Charter School Application Fact Sheet.”)

Alpha School parents and students tell their own stories:

• Austin Scholar. A newsletter by an Alpha School graduate who is very positive about the program. Of particular interest are the articles “Debunking the biggest myths about Alpha,” and “The Alpha App Stack.”
• “Your Review—Alpha School. An anonymous review (mostly positive) of Alpha School by a parent of students at the Alpha School Austin campus.
• JeskaLuv Substack. A newsletter by a parent of students at the Alpha School Brownsville campus, with four posts thus far reviewing their experience (overall negative).
• “Alpha School and 2x Learning.” An article by an Alpha School parent questioning its method for assessing learning improvements.
• “Why I Pulled My 13-Year-Old Daughter Out of Traditional School“ by Rachelle Starr. A (mostly positive) article by a parent whose daughter is a student using the Alpha Anywhere offering for homeschoolers.

Other peoples’ takes on Alpha School and related initiatives:

• “A.I.-Driven Education: Founded in Texas and Coming to a School Near You” by Pooja Salhotra for the New York Times. An overview of Alpha School, including comments from educators and Alpha School parents.
• “AI school offers challenge to public education: ‘They need to be accountable to their customer’” by Zachary Suri for Texas Standard. A story about Alpha School by a Texas-based news organization that interviewed Alpha students (apparently with the cooperation of Alpha PR staff) but was subsequently denied permission to use the interviews, supposedly because the reporter asked too many “DEI questions.”
• “The Truth About 2 Hour Learning & Unbound Academy a/k/a The School ‘Replacing Teachers with AI’” by Dan Meyer. A critical look at the proposed Arizona charter school Unbound Academy, based on Alpha School principles and technology.
• “AI Schools, the Future of Education?” by Quinton Ashley. A critical look at Alpha School and the various people and companies associated with it.
• “Alpha School’s Other Story Wasn’t in the Times” and “Innovation Without Critique Isn’t Reform. It’s Marketing“ by James O‘Hagan. Two critical articles, the first based on the posts in JeskaLuv Substack.
• “A World-Class Education in Two Hours a Day?” by Greg Campion. A positive take on what adoption of the Alpha School model might mean.
• “A New Vision for School” by Joey Kang. Thoughts on Alpha School from a teacher in Singapore.
• “Are Teacherless Schools Good for Education?” by Jordan Davis. Another take from a teacher.
• “What Traditional Education Can Learn from Alpha School’s AI Teaching Model” by David Hobbs.
• “AI’s Impact on Education: High Quality Learning in Just Two Hours a Day“ by Dadang Irsyam. One of the earlier articles on Alpha School, by a tech journalist.
• “On Alpha School” by Zvi Mowshowitz. An positive article discussing some of the principles behind Alpha School, including motivation and gamification.
• X post by Rui Ma. A recap of the AstralCodexTen review, interesting for the subsequent thread.
• “Texas Businesswoman Wants to Open AI-Driven, Teacherless Cyber Charter School in Pennsylvania” by Peter Greene for the Bucks County Beacon. An article criticizing Unbound Academy’s proposal for a charter school in Pennsylvania.
• “RE: Unbound Academic Institute (PA) Charter School.” The Pennsylvania Department of Education’s formal decision rejecting Unbound Academy’s charter school application.