Wealthy US families are increasingly turning away from traditional schools and instead sending their children to AI-powered private schools. As Wall Street Journal reporter Katherine Bindley reports, parents justify this by stating that AI will fundamentally change the economy and that traditional learning methods no longer adequately prepare students for it.
The Alpha School: two hours of AI instruction, then project work
Exemplifying this trend is the Alpha School, founded about twelve years ago in Austin, Texas. The concept: core subjects are taught for two hours daily using AI-powered software that tracks how attentive students are and individually adapts the learning material. The rest of the school day consists of project-based workshops and practical activities. Instead of teachers, the school employs so-called “Guides,” who, according to spokesperson Anna Davlantes, are all paid six figures. Tuition ranges from $10,000 to $75,000 per year, depending on the location.
The school is currently expanding rapidly: according to WSJ, eight new locations were added in 2025, including in San Francisco and New York, with nearly two dozen more expected in the fall of 2026. According to Davlantes, many New York Alpha families work in the finance sector or are entrepreneurs, while in the Bay Area, they predominantly come from the tech industry. One of the most prominent advocates is billionaire Bill Ackman.
Who sends their children there – and why
San Francisco-based venture capitalist Shaun Johnson wants to send his son to the Alpha kindergarten after being dissatisfied with the public school his family was assigned to by lottery. He told the Wall Street Journal that the education system is probably broken, and there will be entrepreneurs trying to fix it. The key factor in his decision was primarily the personalization through AI, not the use of technology for its own sake.
Critical voices: Homogeneity and the role of the teacher
Not all observers view the model uncritically. Victor Lee, a professor at the Stanford Graduate School of Education, expressed concern to the Wall Street Journal that a profile heavily focused on entrepreneurship or AI could lead to a more homogeneous student body than at traditional private schools. Additionally, the avoidance of the term “teacher” contributes to diminishing the role and professionalism of the teaching profession. Structural questions also remain open: According to the Wikipedia entry on the Alpha School, the school’s claims that children learn twice as much as at traditional schools are based on internal analyses by the school itself and have not yet been independently verified. The corporate structure surrounding several interconnected profit-oriented suppliers of the school has also faced criticism.
Two studies show: The traditional system struggles with AI
That traditional educational institutions appear overwhelmed by the use of AI in teaching is demonstrated by two recent studies. A study by the CEPR (Centre for Economic Policy Research) using panel data from 26,811 Chinese secondary school students over 30 months found that while generative AI improved homework grades by 18 percent and reduced processing time by about 30 percent, the results in closed exams dropped by 20 percent within six months. For particularly important entrance exams, the decline after about two years ranged from 18 to 24 percent. About 80 percent of the affected students exhibited a behavior pattern indicating pure outsourcing of cognitive work to AI, according to the study.
A study by the Center for Studies in Higher Education at UC Berkeley reached a similar conclusion, analyzing more than 500,000 grade records from 2018 to 2025 at a selective public university in Texas. In courses with a high proportion of writing and programming tasks, the share of top grades increased by 13 percentage points after the introduction of ChatGPT in November 2022, corresponding to an increase of about 30 percent compared to the baseline. The effect was particularly evident in unsupervised homework, but not in oral presentations – a pattern that, according to study author Igor Chirikov, suggests outsourced work rather than genuinely improved learning performance.
Educational inequality in the age of AI
With annual fees of up to $75,000, a model like Alpha School remains reserved for a small, wealthy minority. At the same time, practically everyone with internet access today potentially has a patient, individually adaptable, and available 24/7 AI tutor – a potential for educational equity, the utilization of which requires the very competence to use AI sensibly rather than as a substitute for one’s own thinking. Whether and how this competence is taught remains largely open at most traditional schools.
Context
The boom of AI-powered private schools among wealthy US families can be seen as a reaction to a real uncertainty about how education should change in light of generative AI – the studies from China and the US show that unreflective use of AI in learning can indeed bring measurable disadvantages. Whether models like Alpha School actually solve this problem or merely hide it behind higher tuition fees and unverified success promises cannot be conclusively assessed based on the currently available, predominantly school-owned data.
