Alpha School, the private school network that charges up to $75,000 a year on the promise that AI tutors let children learn twice as fast, is not hitting that promise for most of its own students. Internal data first reported by The New York Times shows just 42 percent of Alpha students met the network's goal of learning twice as fast in both math and reading.
The number lands at an awkward moment. Alpha is pushing into public schools, where districts are adopting its software hoping to fit more learning into less time, and the company has accumulated powerful backers, from LinkedIn cofounder Reid Hoffman to hedge fund manager Bill Ackman. US Education Secretary Linda McMahon toured the Austin campus in September 2025 and called the model "the most exciting thing I've seen in education in a long time."
How the model works, and where it falls short
Alpha has no traditional teachers. It employs "guides" who do not plan lessons or grade homework, and its core product is TimeBack, an app marketed as an AI tutor that personalizes academic work into two hours a day, freeing the rest of the schedule for workshops on public speaking and other soft skills. Cofounder MacKenzie Price has sold that structure aggressively, and the network has grown to roughly 1,500 students across about 50 campuses.
The math targets are the bigger gap. About half of K-8 students at Alpha's private campuses missed the goal of learning math twice as fast, and roughly one in five missed their reading targets. Students who started behind fared worst: about a fifth of private school students scored at or below the 50th percentile nationally in math last fall, and only 37 percent of that group reached the two-times goal by winter. Guides and parents also told Wired that students worked late into the night to meet the targets. One parent of a 13-year-old reading at an 11th-grade level described reaching them as coming with "stress and anxiety."
The public school test is harder
More than 2,600 public school students in Texas and South Carolina now use Alpha's platform, per company documents reviewed by the Times. Public districts cannot remove struggling students the way a private school can, and they lack the prize incentives Alpha uses at its own campuses. The Houston Independent School District, the largest in Texas, is piloting the software at two schools; Superintendent Mike Miles told the Times that students spent more time on the apps than planned, cutting into other activities, though he still wants AI in the district's toolkit.
Complaints from inside are documented too. Former employees told 404 Media that Alpha's AI generated faulty lesson plans, which internal documents said sometimes did "more harm than good," and that the company scraped other online courses without permission.
What the evidence outside says
Independent research on AI tutoring is not kind to the two-hour premise. A study of 26,811 secondary students in central China found homework scores rose 18 percent after students adopted generative AI tools while closed-book exam scores fell 20 percent, a pattern that looks like help with the answer rather than learning. The OECD's latest PISA results found daily AI users scored lower in science than nonusers, with moderate users about even.
None of this settles the question of whether AI tutoring can work; it does suggest the burden of proof sits with the vendor, not the parent writing the tuition check. Schools are already negotiating with AI vendors under pressure, and Google's threat to switch off Search for districts that blocked its AI answers showed how coercive that market can get. Even tools that work as advertised carry tradeoffs, as the fight over AI reviewers bottlenecking Linux patches demonstrated in software. Alpha's own numbers, released under pressure rather than by choice, are now the best available measure of its central claim, and they fall well short of it. The expansion into public schools makes that a taxpayer question, not just a private one.