Reinventing School for the AI Era | Joe Liemandt, Gagan Biyani & Erik Torenberg

Alpha’s model replaces the conventional teacher-fronted day with a learning engine that adapts material to each student’s level and allows them to progress when ready. Joe Liemandt says students can learn in two academic hours what usually takes six classroom hours plus homework, and that the system aims to keep them around 80–85% correct; below roughly two-thirds success, he says, students risk disengaging. Alpha’s three commitments are that students love school, learn more in two academic hours, and use the rest of the day to develop life skills. In recent surveys, more than 90% said they loved school, and 50% said they would rather attend than go on vacation, rising to 65% at one campus.

Afternoons at Alpha focus on skills such as leadership, teamwork, entrepreneurship, financial literacy, relationships, public speaking, and grit. Liemandt argues that high standards can make students happier when they are tackling difficult challenges with friends. The school treats these capabilities as deliberately as academic subjects: he says third-graders used team-oriented language in fewer than 20% of interactions before an eight-week workshop, which requires 85% team behavior to pass. Teams finish the escape-room exercise three times faster when they collaborate. He also says fourth- and fifth-graders must pass Wharton’s Mount Everest teamwork challenge, which 50% of MBA participants fail.

The speakers argue that grades and classroom instruction alone do not prepare students for work that increasingly requires maturity, ownership, systems thinking, leadership, teamwork, and the ability to receive feedback. Projects and real responsibilities can make learning more motivating, especially when students build something useful, earn revenue, or secure a job. More than 70 MIT and Stanford students who interned at Alpha said they feared university would not prepare them for a world where AI replaces entry-level jobs. They proposed a free educational LLM for worldwide use, including small models that run on devices and separate educational models for each sovereign, and said they wanted university to focus more on collaborating to build difficult things with real-world impact.

Liemandt says AI can help students acquire domain knowledge roughly 10 times faster, potentially reaching master’s-level knowledge in a year, and let them learn while building. He says high-school students are already rebuilding software he wrote and predicts students will write most learning software within two years, with peers learning from one another’s work. He also cites a student-built AP curriculum in which every student who took it earned a 5; its World History course used a game and a snarky AI. The advice to students is to pursue fields that genuinely interest them and become AI-native in those fields, rather than choosing mainly for perceived protection from AI. The speakers expect intelligence to become less scarce, making relationships and practical building skills more valuable.

The speakers see the conventional school day as a design problem, not one that can be solved simply by inserting an AI learning engine into an unchanged schedule. Liemandt expects schools, universities, and education ministries to reconsider the day once they accept that traditional curricula can be learned much faster. Biyani cautions that K–8, high school, and college need different designs, while Liemandt argues that universities need reinvention without expecting people to stop attending. Alpha High graduates praised university seminars, clubs, and activities but described large introductory lectures as a waste of time; one speaker estimated attendance at lecture classes had dropped from about half full to well under 10%. Liemandt says university leaders often spend 90% of their time on adults around the institution rather than student success.

Liemandt frames the shift as comparable to the move from farm labor to tractors: he says more than 30% of Americans worked as farmers when tractor advertisements claimed one machine could do the work of 50 farmhands, and argues that an industrial-era school system now needs to change. Alpha grew from 12 campuses last year to 50, but he says more proof and scale are needed to persuade families that students can learn what they need in two hours. A New York City campus that began with 10 or 12 students is now full at about 150, while some early campus launches struggled to enroll 25 families. He notes that new schools have hiccups and that families opposed to AI or screen time may not be a fit. The broader goal is to raise optimistic, high-agency builders who can use AI to shape the future, rather than train students only for jobs that may change.