Scott Alexander, of Slate Star Codex / Astral Codex Ten fame, ran an Everything-Except-Book Review Contest 2025 in February. The prompt: “Submit an ACX-length post reviewing something, anything, except a book.” The submissions were collected anonymously in a giant 450-page Google Doc. I don’t think the winners have been chosen yet, but there is one essay that has been making the rounds and getting shared more, and that’s Alpha School and “2-hour Learning” powered by AI, a parent’s perspective on Alpha School, a set of “AI-powered” schools in Texas and Florida.
It’s worth reading the entire essay, but I wanted to excerpt a few points I found interesting:
After twelve months I’m persuaded that Alpha is doing something remarkable—but that almost everyone, including Alpha’s own copywriting team, is describing it wrong:
- 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
…Yet the core claim survives: Since they started in October my children have been marching through and mastering material roughly three times faster than their age‑matched peers (and their own speed prior to the program).
One of the surprises doesn’t come until Part 4 of the essay:
Incentives
People REALLY don’t like the idea of incentivizing kids to learn.
Roland Fryer, who has done extensive work on what works in incentivizing students, quotes a 2010 Gallup poll that found that only 23% of American parents support the “idea of school districts paying small amount of money to students to, for example, read books, attend school or to get good grades” (76% opposed the idea with only 1% undecided).
There are not many things that 76% of Americans agree on. Only 69% of Americans believe another Civil War would be a bad thing. Only 78% agree that American independence from Britain was the right choice. People REALLY don’t like paying kids to read books.
I hope that gives you enough of a hook to read the entire essay, it was quite good and provocative to many assumptions I’ve had about education.
Great essay. It’s sad, but I could tell exactly where this parent sent their kids to school previously, just based on the conversations I had with their admissions staff when we were doing our own nation-wide search. (“No acceleration? You’re joking, right?”)
There are decent gifted-only private schools out there that do proper acceleration – alongside same-age peers whenever possible! – but you really do have to dig for them.
This reminds me of a [Veritasium video/presentation](https://youtu.be/0xS68sl2D70?si=cT8R6PuIuj573BZV) I watched about a month ago on AI and Learning. The video is convincing enough for me, ie AI is a great tool for education, but isn’t about to revolutionize it.
why do you think that? i’ve heard the same sentiment from several educators, then i look at stats that show AI tools have been adopted in the classroom faster than any technology and think we might just be ignoring the reality: AI has already revolutionized education.
Because “revolutionize” carries a very strong sentiment, and I don’t think this is the case here – at least not yet. The presentation/video concluded what makes kids learn is a lot more than just the tools used to help in that education.
As a tool, sure, AI is the most revolutionary to date, and has already been shown to have higher efficacy than traditional methods. It’s probably also the most accessible, making it faster to adopt. However, putting someone in front of a screen or device, regardless of interactivity level, doesn’t replace all of the other bits needed to secure a deep understanding of a topic.
Thanks for sharing! We definitely need more alternative approaches to teaching.
Having taught programming and other computer science courses for many years, I find it very disruptive that we have to split the content into chunks to fit the number of arbitrary semester weeks. I achieve better results when I teach the entire syllabus in a 2–3 day, 3–4 hour bootcamp right before the semester begins. The semester can then be used to work on practical projects.
Love the emphasis on mastery and that no topic is left behind (unmastered topics being revisited later). It would be great if this could be incorporated in a university setting. Needing only to know about half the content to pass a course leaves many, MANY blind spots.
thanks matt! giving time is indeed the most important pedagogical tool i’ve found during 15 months as “Father.” my daughter has a larger vocabulary than her peers, largely because we engage her 24/7. she speaks the words we taught her and there’s rarely a moment when she’s not conversing with family.
still, i’m hopeful that AI tools can provide 1:1 support for kids whose family can’t afford to engage 24/7.
the most insightful note i found in that article: “the average tutored child improved by two standard deviations over the control”