This section reveals a lot about the difference between the hype and the practice:
> It isn’t genuine two‑hour learning: most kids start school at 8:30am, start working on the “two-hour platform” sometime between 9am-930am and are occupied with academics until noon-1230pm. They also blend in “surges” from time to time to squeeze in more hours on the platform.
> It isn’t AI in the way we have been thinking about it since the “Attention is all you need” paper. There is no “generative AI” powered by OpenAI, Gemini or Claude in the platform the kids use – it is closer to “turbocharged spreadsheet checklist with a spaced‑repetition algorithm”
> It definitely isn’t teacher‑free: Teachers have been rebranded “guides”, and while their workload is different than a traditional school, they are very important – and both the quantity and quality are much higher than traditional schools.
> The bundle matters: it’s not just the learning platform on its own. A big part of the product’s success is how the school has set up student incentives and the culture they have built to make everything work together
So in other words, they're trying to set up a generally high quality education system, but they have a marketer on board who knows how to capture headlines with controversial claims?
It sounds like they are trying to replicate Asian style rote learning and cram school in a way that's more palatable to western audiences. Rebranding rote learning and force feeding as spaced repetition and surges. Typical SV bull.
They're copying one of Khan Academy's implementation models [1] and rebranding it as AI. It's certainly not new besides the "help yourself to AI" part (which, full disclosure, Khan Academy is working on as well with their "Khanmigo" assistant [2]). Sal Khan, the founder and CEO of Khan Academy, did a TED talk [3] on this.
The scrollbar is completely absent in Firefox. I think this is the first time I've seen long-form content with zero visual indication where I currently am in the document. Crazy.
Edit: Actually the scroll-bar is there, but it's nearly impossible to see because of the low contrast with the background. I guess I can blame my user agent for this one.
I really, really hate the modern trend of scrollbar design. I guess it makes some amount of sense if you're aiming for a mobile phone factor, where real estate is somewhat limited, but changing the scrollbar from a widget that lives functionally outside of the content it is scrolling to a translucent show-only-on-hover widget that overlays the content (and can thus become functionally invisible if the content is just the wrong color) is a real step backwards in UI design.
keep in mind this piece was 8 months ago (and probably starting to be written much earlier). I can see Alpha school being this genuine effort to offer an alternative education plan as well as slowly falling into the AI hype later on and pushing more generative content as it tries to phase out teachers.
Or, if it's being praised by this administration, it's doing so to gain political points.
Holy crap that is a long article. In my view, the only important point is time freed up which should be a part of normal education. If students had more free time to think and contemplate one wonders what kind of world we would live in.
Too bad it takes a dubious idea for an AI school to surface that wisdom.
>the only important point is time freed up which should be a part of normal education.
College felt like the last time I truly had "time freed up". 16 units of classes per quarter came down to ~8-10 hours of class time per week and the recommendation was 2 hour of study per class unit a week (e.g. a 4 unit class recommends 8 hours of study a week). So, typical full time work. But mostly on campus (aka a walkable community), close to peers, with no worries of future responsibilities.
now of course, the CS curriculum easily required double or triple that recommendation, but that speaks more towards the subject than the concept of college.
Well people who are evangelically praising AI for writing code seem to be converging on established, well-known software practices for "making the AI work better" so seems like AI is great at making people rediscover the practices they didn't feel like putting effort into before AI.
AI is held to a much higher standard than the existing education techniques.
Even without AI teachers are implementing new techniques without any evidence of their effectiveness. In some cases, there is mountains of evidence that their techniques are not effective. From the prohibition on phonics in reading, learning styles, building thinking classrooms, or just the entire idea of constructivism. These are all worse than the techniques that they replace. AI systems at the very least are measured and have some kind of tracking of what works.
I'm not advocating for AI necessarily, but education is in the pre-scientific phase and it needs to start by implementing evidence based techniques, AI or no.
I am both pro and against this at the same time. I love the idea of tracking it as an aggregate, but I hate the idea that the kid may end up being stuck on some vibe coded idiocy and unable to move on, because of it ( I still shudder at some of the ridiculous math tests in college that could not account for the right answer, but not in the exactly right format that was not disclosed as expected ).
I am not even suggesting in person teaching is the only solution either. I am currently dealing with, apparently, my kids teacher, who, well, kinda checked out, but as much I am happy for her being able to retire soon, I am not sure why my kid has to suffer academically.
What I am saying is, there is room for AI. What I worry about is, people are idiots and anything half-useful will be neutered and kids will have all the drawbacks of heavy surveillance and zero to show for it.
First let's stick to the lower grades (under 5th). The evidence isn't as clean for upper grades.
Constructivist teaching favors things like student-centered discovery, inquiry-based, minimal-guidance, "child-led" or "whole-language" approaches.
This is just a plain bad way to teach the basics, like reading, writing and arithmetic. People didn't just magically invent these ideas. Most of human history is pre-literacy. Why are we expecting a 5yo to spontaneously learn to read?
This has been studied extensively. Have a look at Project Follow Through (1968–1977). It's the largest study of its kind.
This U.S. government-funded study involved ~120,000 disadvantaged K–3 students across 20+ instructional models in multiple sites. It directly compared Direct Instruction (Engelmann’s scripted, explicit basic-skills model, e.g., DISTAR) against constructivist-oriented models (e.g., Bank Street child-directed, Open Education/EDC exploratory-discovery, cognitive-conceptual discovery approaches).
Abt Associates did an independent evaluation in (1977). Their findings
DI produced the highest gains in reading, math, language, and spelling—raising performance to near national averages. It was the only model with consistent positive effects across basic skills, cognitive-conceptual skills, and self-esteem/affective outcomes.
For more recent evidence, have a look at the reforms in Mississippi and the UK. Mississippi has striking gains for under privileged students.
Mississippi Columbus Municipal School District used the "Reading Mastery Signature Edition" DI program.
Demographics: 92% African American, 100% free/reduced-price lunch, 12% special education.
Results as measured by NWEA MAP and Renaissance STAR assessments:
MAP RIT gains: +15.96 points overall
43–45% of students met or exceeded expected growth; top performers gained ~20–28 points.
Similar results in:
Baltimore City Public Schools (1996–2008)
Arthur Academies (Portland, Oregon metro, 2007–2013)
Rimes Elementary, Florida (2011)
Gering Public Schools, Nebraska (2004+)
> “All educational content is obsolete. Every textbook, every lesson plan, every test, all of it is obsolete because gen AI is going to be able to deliver a personalized lesson just for you,” Joe Liemandt, Alpha School’s “principal” and the founder of Trilogy, the company that owns many of the apps used by Alpha School, said in a podcast interview published last year.
I wonder if this fellow has ever read a serious book. I'm skeptical.
I'm highly skeptical of someone bragging about barely reading a book year to then say AI is going to disrupt education as an authoritative voice. Do you honestly know any teachers or have children in public education? These takes are truly baffling. It's right up there with believing InfoWars as sound politics.
edit: to try and sound "nicer," would anyone seriously take any advice about software engineering from someone who uses a WYSIWYG editor? That's how the above comment reads.
>and it is infinitely better than a book or a teacher.
why do you think that?
>The student could move at his/her own pace and can ask questions if stuck which no book or teacher could deliver.
You're assuming there's a driven student who already knows what they wish to pursue. Even in college I wasn't entirely sure what domain in tech I wanted to explore.
You're also assuming or dismissing the other factors a teacher offers. Networks, parental guidance, wisdom of how to navigate through college or a job market, or simply as emotional pillar in ways parents can't always be.
Most of all, teachers teach you to understand bias and expand your viewpoints past any one given source. That's why I read the parental review of this school on top of this piece. It seems against current coporate interests to offer that. There's no one clear answer to everything out there, but AI wants it to appear as so.
> Around 20 non-fiction and technical books in about 15 years
That's not lot, mate. Maybe more than the average American unfortunately, but I consider a year where I get through 2–3 books a slow one. And that's just reading I do outside of my job. What matters, of course, is both the quality and quantity of what you read. The short of it is, your attempt at building your ethos has fallen pretty flat here.
> AI is going to disrupt the whole academia [sic]
Yes, it has and it continues to. I'm not arguing against that. Joe Liemandt said that "all educational content is obsolete", which presumably includes not only textbooks like SICP or Sipser's Introduction to the Theory of Computation (just listing some CS textbooks because I'm in CS) but also great works of literature and philosophy that are important texts like The Odyssey or A Tale of Two Cities. If he meant to exclude such texts from the umbrella of "all educational content", well, then that's telling too. :)
> …it is infinitely better than a book or a teacher.
Maybe for someone who struggles with literacy or who hasn't had the pleasure of a good teacher. If you really believe this I'd like to see you try to substantiate your claim.
> The student could move at his/her own pace…
The article is about a grade school kids who, most of the time, need a little pushing to reach their full potential.
> …ask questions if stuck which no book or teacher could deliver.
… you're saying that LLMs are better than teachers because you can ask LLMs questions and not teachers?! Also, asking questions isn't the only component of learning. A good teacher will know when to not answer a question (or ask one!) and let the student stew and think about it.
I'm not saying AI can help with education. It can—it helps me!—but no hallucinating stochastic machine will have the human insight that a good teacher has. It's not a replacement.
I still find books valuable because they give you a structure and physical location to anchor learning. They give you an overview of the whole topic you're interested in.
Whenever I need to learn a new topic I always try to buy a book on the subject because it so much faster to have someone do the scaffolding for you than trying to be 'self-directed' about it. It also give some confidence that you aren't leaving big holes in your understanding.
What? The textbook+teacher combo literally provides exactly that.
The textbook allows you to move at your own pace, acting as a structured reference and practice tool that you can review endlessly outside of class.
And the teacher can answer any questions you've confirmed you're not able to resolve on your own with the textbook. Some in class, some during office hours/before or after class, and some via email.
I have long conversation threads on highly specialized topics and I’ve never found learning about something so easy. He’s right but it has to not be self directed in some way because that takes motivation and you cant expect that from every student.
Is anyone shocked that the founder of Trilogy Software, Joe Liemandt, would take the lessons learned in creating a "bossware" enterprise software stack (Crossover) and apply them to his latest venture?
I think there is enormous opportunity in combining Ed Tech and Generative AI, especially if you can create a highly tailored tutor available 24/7 for every student, especially for those in low-income situations that have historically been locked out from gaining such guidance. It's just unfortunate that this so rapidly morphed to spying on students for data harvesting.
The follow on effects if this becomes pervasive could be so incredibly damaging for students. Anxiety from performance metrics are already a very real thing because of standardized testing and scoring to get into the best schools. Also, imagine all of this data "follows" a student as they transition into the work force. We're headed towards a future where entry-level employees will have to disclose their "course work engagement" KPIs on their resume.
The dream is clearly asking a student what they’re interested in, getting them to self direct on a project with guidance and deliver something you can evaluate. I would hazard that AI helps rather than hinders this.
hot take: if all models stopped improving today, we could get there with today's technology.
No, you couldn't whip out a foundational model out of the box to do this, but we could likely engineer our way to guardrails to support this without substantial hallucinations (i.e. fewer than the 95th percentile of human teachers.)
Also in no way would this replace everything teachers do. I'm talking purely academic curiosity.
Readit News
> It isn’t genuine two‑hour learning: most kids start school at 8:30am, start working on the “two-hour platform” sometime between 9am-930am and are occupied with academics until noon-1230pm. They also blend in “surges” from time to time to squeeze in more hours on the platform.
> It isn’t AI in the way we have been thinking about it since the “Attention is all you need” paper. There is no “generative AI” powered by OpenAI, Gemini or Claude in the platform the kids use – it is closer to “turbocharged spreadsheet checklist with a spaced‑repetition algorithm”
> It definitely isn’t teacher‑free: Teachers have been rebranded “guides”, and while their workload is different than a traditional school, they are very important – and both the quantity and quality are much higher than traditional schools.
> The bundle matters: it’s not just the learning platform on its own. A big part of the product’s success is how the school has set up student incentives and the culture they have built to make everything work together
So in other words, they're trying to set up a generally high quality education system, but they have a marketer on board who knows how to capture headlines with controversial claims?
[1] https://en.khanacademy.org/khan-for-educators/
[2] https://www.khanmigo.ai/
[3] How AI Could Save (Not Destroy) Education | Sal Khan | TED - https://www.youtube.com/watch?v=hJP5GqnTrNo - May 1st, 2023
Summary: the '2× learning' claims are overblown.
Edit: Actually the scroll-bar is there, but it's nearly impossible to see because of the low contrast with the background. I guess I can blame my user agent for this one.
Usual rambling from a site that disables scrolling and nearly crashes Firefox.
Or, if it's being praised by this administration, it's doing so to gain political points.
Too bad it takes a dubious idea for an AI school to surface that wisdom.
College felt like the last time I truly had "time freed up". 16 units of classes per quarter came down to ~8-10 hours of class time per week and the recommendation was 2 hour of study per class unit a week (e.g. a 4 unit class recommends 8 hours of study a week). So, typical full time work. But mostly on campus (aka a walkable community), close to peers, with no worries of future responsibilities.
now of course, the CS curriculum easily required double or triple that recommendation, but that speaks more towards the subject than the concept of college.
Even without AI teachers are implementing new techniques without any evidence of their effectiveness. In some cases, there is mountains of evidence that their techniques are not effective. From the prohibition on phonics in reading, learning styles, building thinking classrooms, or just the entire idea of constructivism. These are all worse than the techniques that they replace. AI systems at the very least are measured and have some kind of tracking of what works.
I'm not advocating for AI necessarily, but education is in the pre-scientific phase and it needs to start by implementing evidence based techniques, AI or no.
I am not even suggesting in person teaching is the only solution either. I am currently dealing with, apparently, my kids teacher, who, well, kinda checked out, but as much I am happy for her being able to retire soon, I am not sure why my kid has to suffer academically.
What I am saying is, there is room for AI. What I worry about is, people are idiots and anything half-useful will be neutered and kids will have all the drawbacks of heavy surveillance and zero to show for it.
Constructivist teaching favors things like student-centered discovery, inquiry-based, minimal-guidance, "child-led" or "whole-language" approaches.
This is just a plain bad way to teach the basics, like reading, writing and arithmetic. People didn't just magically invent these ideas. Most of human history is pre-literacy. Why are we expecting a 5yo to spontaneously learn to read?
This has been studied extensively. Have a look at Project Follow Through (1968–1977). It's the largest study of its kind.
This U.S. government-funded study involved ~120,000 disadvantaged K–3 students across 20+ instructional models in multiple sites. It directly compared Direct Instruction (Engelmann’s scripted, explicit basic-skills model, e.g., DISTAR) against constructivist-oriented models (e.g., Bank Street child-directed, Open Education/EDC exploratory-discovery, cognitive-conceptual discovery approaches).
Abt Associates did an independent evaluation in (1977). Their findings
DI produced the highest gains in reading, math, language, and spelling—raising performance to near national averages. It was the only model with consistent positive effects across basic skills, cognitive-conceptual skills, and self-esteem/affective outcomes.
See page 311 fig1 for the plot of outcomes. https://andymatuschak.org/files/1988-Engelmann.pdf
For more recent evidence, have a look at the reforms in Mississippi and the UK. Mississippi has striking gains for under privileged students. Mississippi Columbus Municipal School District used the "Reading Mastery Signature Edition" DI program.
Demographics: 92% African American, 100% free/reduced-price lunch, 12% special education.
Results as measured by NWEA MAP and Renaissance STAR assessments:
MAP RIT gains: +15.96 points overall 43–45% of students met or exceeded expected growth; top performers gained ~20–28 points.
Similar results in: Baltimore City Public Schools (1996–2008) Arthur Academies (Portland, Oregon metro, 2007–2013) Rimes Elementary, Florida (2011) Gering Public Schools, Nebraska (2004+)
Dead Comment
I wonder if this fellow has ever read a serious book. I'm skeptical.
AI is going to disrupt the whole academia and it is infinitely better than a book or a teacher.
The student could move at his/her own pace and can ask questions if stuck which no book or teacher could deliver.
edit: to try and sound "nicer," would anyone seriously take any advice about software engineering from someone who uses a WYSIWYG editor? That's how the above comment reads.
why do you think that?
>The student could move at his/her own pace and can ask questions if stuck which no book or teacher could deliver.
You're assuming there's a driven student who already knows what they wish to pursue. Even in college I wasn't entirely sure what domain in tech I wanted to explore.
You're also assuming or dismissing the other factors a teacher offers. Networks, parental guidance, wisdom of how to navigate through college or a job market, or simply as emotional pillar in ways parents can't always be.
Most of all, teachers teach you to understand bias and expand your viewpoints past any one given source. That's why I read the parental review of this school on top of this piece. It seems against current coporate interests to offer that. There's no one clear answer to everything out there, but AI wants it to appear as so.
That's not lot, mate. Maybe more than the average American unfortunately, but I consider a year where I get through 2–3 books a slow one. And that's just reading I do outside of my job. What matters, of course, is both the quality and quantity of what you read. The short of it is, your attempt at building your ethos has fallen pretty flat here.
> AI is going to disrupt the whole academia [sic]
Yes, it has and it continues to. I'm not arguing against that. Joe Liemandt said that "all educational content is obsolete", which presumably includes not only textbooks like SICP or Sipser's Introduction to the Theory of Computation (just listing some CS textbooks because I'm in CS) but also great works of literature and philosophy that are important texts like The Odyssey or A Tale of Two Cities. If he meant to exclude such texts from the umbrella of "all educational content", well, then that's telling too. :)
> …it is infinitely better than a book or a teacher.
Maybe for someone who struggles with literacy or who hasn't had the pleasure of a good teacher. If you really believe this I'd like to see you try to substantiate your claim.
> The student could move at his/her own pace…
The article is about a grade school kids who, most of the time, need a little pushing to reach their full potential.
> …ask questions if stuck which no book or teacher could deliver.
… you're saying that LLMs are better than teachers because you can ask LLMs questions and not teachers?! Also, asking questions isn't the only component of learning. A good teacher will know when to not answer a question (or ask one!) and let the student stew and think about it.
I'm not saying AI can help with education. It can—it helps me!—but no hallucinating stochastic machine will have the human insight that a good teacher has. It's not a replacement.
Whenever I need to learn a new topic I always try to buy a book on the subject because it so much faster to have someone do the scaffolding for you than trying to be 'self-directed' about it. It also give some confidence that you aren't leaving big holes in your understanding.
The textbook allows you to move at your own pace, acting as a structured reference and practice tool that you can review endlessly outside of class.
And the teacher can answer any questions you've confirmed you're not able to resolve on your own with the textbook. Some in class, some during office hours/before or after class, and some via email.
Deleted Comment
I think there is enormous opportunity in combining Ed Tech and Generative AI, especially if you can create a highly tailored tutor available 24/7 for every student, especially for those in low-income situations that have historically been locked out from gaining such guidance. It's just unfortunate that this so rapidly morphed to spying on students for data harvesting.
The follow on effects if this becomes pervasive could be so incredibly damaging for students. Anxiety from performance metrics are already a very real thing because of standardized testing and scoring to get into the best schools. Also, imagine all of this data "follows" a student as they transition into the work force. We're headed towards a future where entry-level employees will have to disclose their "course work engagement" KPIs on their resume.
Not this AI, today
No, you couldn't whip out a foundational model out of the box to do this, but we could likely engineer our way to guardrails to support this without substantial hallucinations (i.e. fewer than the 95th percentile of human teachers.)
Also in no way would this replace everything teachers do. I'm talking purely academic curiosity.
Deleted Comment