Parents pay $75k for Alpha's AI school. Half of its students missed math

Alpha School expanded its private education network charging up to $75,000 annually while pitching a software-driven curriculum that promises students will learn twice as fast as their peers in two hours of daily academics. According to fortune.com's report, internal records show that only 42% of students achieved that twin benchmark in math and reading across its campuses. The organization operates roughly 50 locations serving 1,500 private school students while simultaneously rolling out its proprietary learning platform into traditional public school districts.

This report examines internal performance metrics, software reliability concerns, and public school expansion data for educators, parents, and district administrators evaluating automated tutoring software. It outlines how algorithmic instruction models perform across different student demographics and analyzes academic studies tracking the measurable impact of machine learning tools on pupil comprehension.

01Two-hour school days and high-profile backers

Alpha replaces certified classroom instructors with adult guides who do not plan lessons, grade student homework, or deliver lectures. Cofounder MacKenzie Price structured the academic day around an application called TimeBack, designed to accelerate instructional delivery and compress core coursework into a two-hour window. The remaining daylight hours are reserved for life skills, public speaking exercises, and interpersonal relationship building.

Enrollment expanded from over 1,200 pupils in September to approximately 1,500 students across 50 locations following ambitious growth announcements in August. Silicon Valley venture capitalist Sarah Cone enrolled her eight-year-old daughter at the New York campus, emphasizing that learning to utilize automated systems has become an essential capability for modern children.

Political and financial figures have endorsed the model publicly, including LinkedIn cofounder Reid Hoffman and hedge fund manager Bill Ackman on X. In September 2025, United States Education Secretary Linda McMahon toured the Austin facility and praised the structure as an exciting educational development. Even the social media account of Theranos founder Elizabeth Holmes shared commendations of Price.

02Internal test data and double-speed targets

Network performance logs revealed significant disparities between marketed claims and standardized student progress benchmarks across network facilities. While forty-two percent of pupils reached the double-speed standard in both core subjects, approximately half of kindergarten through eighth-grade attendees failed to meet the designated mathematics growth velocity. Reading outcomes proved somewhat steadier, though one in five pupils missed targeted reading growth thresholds.

Pupils who began the academic year below national medians experienced the steepest struggles under the automated format. Roughly twenty percent of private school attendees entered the previous autumn testing at or below the fiftieth percentile nationally in arithmetic. By the winter testing interval, merely thirty-seven percent of that lower-scoring cohort attained the target two-times acceleration rate.

03Evening study strain and faulty lesson plans

The core marketing claim that children complete required coursework within two daytime hours encountered contradiction from families and staff. Guides and parents reported that numerous pupils spent additional hours working late into the night at home to satisfy rigorous institutional milestones. One parent of a thirteen-year-old reading at an eleventh-grade level stated that maintaining required performance pace produced noticeable anxiety.

Former staff members disclosed that the underlying systems generated defective educational material, which internal operational logs noted frequently caused more harm than good during active sessions. Personnel also stated that the organization scraped online educational courses without securing authorization from intellectual property owners. These automated generation errors complicated daily workflows for on-site guides attempting to supervise frustrated students.

04Public school rollouts across southern districts

State education agencies and municipal districts expanded software adoption beyond private facilities, placing the platform before more than 2,600 public school students in Texas and South Carolina. The Houston Independent School District, representing the largest public school system in Texas, initiated a multi-campus pilot program across two facilities. Superintendent Mike Miles noted that platform sessions consumed more classroom hours than originally scheduled.

Superintendent Miles maintained that technological literacy remains vital for preparing children for evolving modern workplaces despite early operational friction. Municipal classroom deployments face distinct administrative environments compared to private campuses that can dismiss underperforming students or offer lavish prize incentives to motivate software engagement.

Note: Public school contracts operate under mandatory enrollment regulations where administrators cannot dismiss underperforming children or distribute financial prize incentives common across private campuses. District pilots in Texas and South Carolina must instead integrate the software alongside standard curriculum requirements within diverse general education environments.

05Academic studies on artificial intelligence penalties

Independent educational research highlights persistent cognitive drawbacks associated with automated tutoring and generative systems. A rigorous academic investigation tracking 26,811 secondary students across central China discovered that homework assignment scores rose eighteen percent following tool adoption. However, corresponding closed-book testing scores declined twenty percent among the exact same student cohort.

Global metrics from the Programme for International Student Assessment revealed that pupils who relied daily on automated computational tools registered lower science evaluations than nonusers, whereas moderate users demonstrated steady performance. University professors report that habitual reliance on algorithmic summaries impairs long-term student ability to process extensive text passages.

University of Notre Dame theology professor Timothy O'Malley observed that students assigned substantial volumes of reading frequently struggle to navigate the text independently. Pepperdine University humanities professor Jessica Hooten Wilson warned that heavy reliance on algorithmic shortcuts degrades foundational literacy, characterizing the issue as an inability to read complex sentences.