Key findings
- Alpha School’s headline results are self-reported. The company says students grow 2.6 times faster than peers on MAP tests and “learn 2X in 2 hours,” but those figures come from its own analyses; no independent or peer-reviewed study of Alpha’s outcomes had been published as of October 2026.
- The model: about 2 hours a morning of adaptive learning software for core subjects, afternoon workshops, and adult “guides” instead of traditional teachers. Founders have said the “AI” is mostly adaptive apps, not chatbots.
- The price: $40,000 to $75,000 a year at most campuses listed by Newsweek in September 2026, including $55,000 at the new Chicago campus that opened for 2026-27.
- Regulators are split. Of five states where an affiliated online charter applied, only Arizona approved; Pennsylvania called the model “untested” and denied it in 2025.
- AI tutoring research is mixed, and design decides it. In a peer-reviewed trial of about 1,000 students, an unrestricted GPT-4 tutor lowered later exam scores by 17%; a version with teacher-built guardrails did no harm, and a Harvard trial found a carefully designed tutor beat active learning.
Short answer: Alpha School and similar AI-driven schools promise that two hours a day of adaptive software can replace a full school day of teaching, but their results so far are self-reported. No independent, peer-reviewed study of Alpha’s outcomes had been published as of October 2026. Tuition runs $40,000 to $75,000 a year at most campuses. Peer-reviewed research on AI tutoring shows it can help or hurt depending on design, and none of those trials tested a whole school built on it.
This page separates what Alpha and its affiliates claim from what has been checked by outsiders, lists costs and regulatory decisions, and summarizes the strongest AI tutoring trials. Claims that come from the company are labeled as such.
What Alpha School says it does
Alpha describes a “2-Hour Learning” model: core academics in about two hours each morning on adaptive software, then afternoon workshops. Students work at their own level and pace on apps that adjust as they go. Afternoons cover life skills such as public speaking, coding and financial literacy, according to a September 2026 Newsweek report. Adults called “guides” motivate and mentor rather than lecture.
Alpha was founded in 2014 in Austin, Texas, by MacKenzie Price and Brian Holtz and was formerly called Emergent Academy, according to Wikipedia’s summary of published reporting. The word “AI” needs a footnote: according to the founders, as reported by Chalkbeat, it refers mainly to adaptive learning apps similar to IXL or Khan Academy tools, not chatbots like ChatGPT.
Alpha’s website, quoted by Newsweek, says: “Your kids can accomplish twice as much if they’re not sitting in a one-size-fits-all classroom for 6 hours.” That is a marketing claim, examined below.
Which claims are verified and which are self-reported?
The structure and prices are independently reported. The learning results are not. This table labels each claim by who stands behind it.
| Claim | Who says so | Status |
|---|---|---|
| Core academics take about 2 hours a day | Alpha; described by Newsweek and NBC Chicago | Model description, reported by media |
| Students “learn 2X in 2 hours” | Alpha marketing slogan | Self-reported, no independent test |
| Students grow 2.6 times faster than peers on MAP tests | Alpha, as reported by Newsweek (Sept. 2026) | Self-reported; internal analysis of NWEA MAP data |
| MAP results independently reviewed | Chalkbeat (Jan. 2025), via Wikipedia | Not verified: data had not been independently reviewed |
| Peer-reviewed study of Alpha outcomes | Our search, Oct. 2026 | None found |
| Campuses in 16 states (open, launching or planned) | Newsweek, Sept. 2026 | Independently reported |
| 27 new communities for fall 2026 | Alpha press release, Aug. 2026 | Company announcement |
| Tuition $40,000–$75,000 at listed campuses | Newsweek, Sept. 2026 | Independently reported; prices may change |
Table 1. Alpha School claims by source and verification status, as of October 15, 2026.
Even accurate internal scores would leave a selection problem. Families who pay $40,000 or more for an experimental private school differ from typical public school families in income, motivation and prior achievement. Without a matched comparison group or a lottery, faster growth could reflect who enrolls rather than what the school does.
How much does Alpha School cost?
Most campuses on Newsweek’s September 2026 list charge between $40,000 and $75,000 a year. Wikipedia’s summary of earlier reporting gives a wider range of $10,000 to $75,000 across in-person schools, with most around $40,000.
Source: Newsweek, September 14, 2026. Bars scaled to $80,000. Prices change; check the school directly.
| Tuition | Campuses on Newsweek’s list |
|---|---|
| $40,000 | Scottsdale AZ, Tulsa OK, Keller/Southlake TX |
| $45,000 | Charlotte NC |
| $50,000 | Lone Tree CO, Miami FL, Roswell GA, Park City UT |
| $55,000 | Chicago IL, Kirkland WA |
| $60,000 | Boston MA, Malibu CA, Dallas (Highland Park) TX |
| $65,000 | Armonk NY, Miami Beach FL, Bethesda MD, Lower Manhattan NY, Chantilly VA |
| $75,000 | Palo Alto CA |
Table 2. Alpha School tuition by campus, Newsweek, September 2026.
Worked example. At $55,000 a year, a child who attends Alpha Chicago from kindergarten through 8th grade (9 years) would cost $495,000 in tuition at today’s price, before any increases. If core instruction really takes 2 hours a day over a 180-day year, that is 360 hours of core academics a year, or about $153 per core-instruction hour ($55,000 ÷ 360). The afternoon workshops are part of what families pay for, so this is a framing device, not a full cost comparison.
The Chicago “teacherless” school
Alpha Chicago opened for the 2026-27 school year at 350 E. South Water St. in the Loop, serving pre-K through 8th grade at $55,000 a year. NBC Chicago reported it as the network’s first Illinois campus, with the company operating in at least 15 other states and Puerto Rico. The CBS Mornings segment embedded below this article covered the same opening.
“Teacherless” is shorthand used in headlines. The school has no traditional classroom teachers: software delivers core lessons and adult guides supervise, coach and run afternoon workshops. Founder MacKenzie Price has disputed the framing, saying in local coverage that the school is not replacing teachers and that the role is changing.
Critics quoted in Chicago coverage pointed to the lack of outside research on the curriculum and argued that human teachers provide emotional support and problem-solving skills software cannot. Newsweek also noted that UNESCO has warned about AI in education raising risks for privacy, critical thinking and reduced human interaction.
How have states and regulators responded?
Public approval has been rare. An affiliated organization, Unbound Academic Institute, applied to open online (cyber) charter schools using the model in five states. Wikipedia’s summary of local reporting lists the outcomes.
| State or agency | Decision | Detail reported |
|---|---|---|
| Arizona | Approved | The only approval among the five applications |
| Pennsylvania | Denied (2025) | State education department called the model “untested” and said it did not show alignment with state standards (Pennsylvania Capital-Star) |
| North Carolina, Arkansas, Utah | Applied; not approved | Dates and details not given in the sources we checked |
| Texas Education Agency | Vetoed Valenta Academy charter (June 2025) | Planned Bastrop school using the model was canceled (Community Impact) |
| Texas voucher program | Temporary block, later resolved for most | Schools accredited only by Cognia, including Alpha, were briefly excluded in late 2025 and early 2026 |
Table 3. Regulatory decisions involving Alpha-affiliated schools. Source: Wikipedia, "Alpha School," citing Pennsylvania Capital-Star, Bucks County Beacon, Community Impact, Austin American-Statesman and Houston Chronicle.
The Pennsylvania application, as described by the Bucks County Beacon, planned one head of school and 16 guides (5 senior, 10 junior, 1 special education) for 500 students, with no school building. It budgeted $5,500 per student for 2 Hour Learning, the software provider, compared with $2,000 to $2,500 in other states’ applications. For how public districts are writing their own AI rules, see our tracker of AI policies in 32 large school districts.
What peer-reviewed research says about AI tutoring
Well-designed tutoring software can raise scores; unguided chatbots can lower them. No trial has tested a full AI-led school day. We include only peer-reviewed studies with control groups.
| Study | Design | What was tested | Result |
|---|---|---|---|
| Bastani et al., PNAS (2025) | Field experiment, about 1,000 students in ~50 high school math classes, Turkey | Unrestricted GPT-4 (“GPT Base”) vs GPT-4 with teacher-designed hints and guardrails (“GPT Tutor”) vs textbook only | Practice scores up 48% (Base) and 127% (Tutor); on the later exam without AI, Base scored 17% lower than control, Tutor about the same as control |
| Kestin et al., Scientific Reports (2025) | Randomized crossover trial, 194 students, Harvard introductory physics, fall 2023 | Custom AI tutor built with instructor-written prompts vs an active-learning class | AI group learned more in less time; preliminary analysis put gains at about double |
| Kulik & Fletcher, Review of Educational Research (2016) | Meta-analysis of 50 controlled evaluations | Intelligent tutoring systems (pre-chatbot adaptive software) | Median gain of 0.66 standard deviations; much smaller on standardized than on locally made tests |
Table 4. Peer-reviewed controlled studies of AI and computer tutoring.
Source: Bastani et al. (2025), PNAS 122(26), doi:10.1073/pnas.2422633122. Bar length shows size of change, scaled to 127%; red marks a decline.
The lesson is that design matters more than the label “AI.” The authors concluded students used unrestricted GPT-4 as a crutch: practice looked better while learning got worse. The Harvard tutor (Scientific Reports, 2025) worked because instructors scripted it to guide rather than answer. Alpha’s apps are closer to the older adaptive-tutoring category, where the 2016 meta-analysis shows real but test-dependent gains. Our guide to AI tutors for students compares the tools families can try at home, and our roundup of AI study tools covers flashcard and quiz apps built on the same adaptive idea. Chatbots also make confident mistakes, which is why the AI hallucination problem matters for any school that leans on them.
How students already use AI for school
AI study help is already normal for teenagers, at any school. In a Pew Research Center survey of 1,458 US teens (fall 2025, published February 2026), 54% said they had used chatbots for help with schoolwork, and about 1 in 10 did all or most of their schoolwork with chatbot help (Pew). An earlier Pew question found 26% of teens had used ChatGPT for schoolwork in 2024; the wording differs, so the two are not a clean trend.
PISA 2025 found about half of 15-year-olds across OECD systems used AI at least weekly to help them learn, and heavier use of AI for writing was generally associated with lower reading scores, Education Week reported. That is a correlation, but it lines up with the Bastani trial: shortcuts during practice can cost learning later.
For families, the practical question is less “AI school or not” than “does my child use AI to think or to skip thinking?” Our guides on using AI honestly for schoolwork, whether AI homework help counts as cheating and safe AI tools by age cover that at home. If you want to set family rules in writing, start from our parent letter about AI use, and see the step-by-step guide to using AI for homework help the right way. More on AI and school is collected in our AI hub. Students who rely on screen-based learning for hours should also read our research on student attention span.
Questions parents should ask an AI school
Ask for evidence an outsider can check, and for the details marketing leaves out. Bring this list to a tour.
- Outcomes: Has anyone outside the company reviewed your test results? Can I see growth data by grade, with how many students were tested and how many left?
- Comparison: Growth compared with whom? National norms, or similar students who applied but did not enroll?
- Adults: What is the adult-to-student ratio in the morning session? What training or credentials do guides have?
- Software: Which apps deliver math and reading? Are they the same tools available elsewhere, and what does the school add?
- Support: How do you serve students with IEPs, 504 plans or reading difficulties?
- Data: What student data do the apps collect, who stores it, and can I delete it?
- Transfers: What is your accreditation, and how do credits and records transfer to public schools or state programs?
- A real day: Can I observe a full morning, not just a demo?
A short version for any school, AI-driven or not, is in our parent-teacher conference questions.
Methodology and limitations
We compiled published reporting and peer-reviewed studies; we did not visit a campus or test the software. Facts were checked on October 15, 2026.
- Company claims (2.6× growth, “2X in 2 hours,” 27 new communities) are reported as claims, with the source named. We found no independent data to confirm or refute them.
- Several regulatory details come from Wikipedia’s summary of local news coverage; we cite the original outlets named there.
- Tuition is from Newsweek’s September 2026 list and may change.
- Research scope: we included only peer-reviewed, controlled studies. A widely shared World Bank trial in Nigeria was left out because it was published as a working paper and blog, not a peer-reviewed article.
- Neutrality: we have no relationship with Alpha School, 2 Hour Learning or any school named here.
Cite this page
Sources
- Newsweek, “Map shows where teacher-free schools are expanding across America,” September 14, 2026: newsweek.com
- NBC Chicago, Alpha Chicago opening, September 9, 2026: nbcchicago.com
- Wikipedia, “Alpha School” (summarizing Chalkbeat, Pennsylvania Capital-Star, Bucks County Beacon, Community Impact, Austin American-Statesman, Houston Chronicle): en.wikipedia.org
- Alpha School press release on 27 new communities, August 2026 (company announcement): pr.davisjournal.com
- Bastani, H. et al. (2025). Generative AI without guardrails can harm learning: Evidence from high school mathematics. PNAS 122(26): doi.org ; summary: chibe.upenn.edu
- Kestin, G. et al. (2025). AI tutoring outperforms in-class active learning: an RCT introducing a novel research-based design in an authentic educational setting. Scientific Reports 15, 17458: nature.com ; Harvard Gazette (2024): news.harvard.edu
- Kulik, J. A. & Fletcher, J. D. (2016). Effectiveness of intelligent tutoring systems: A meta-analytic review. Review of Educational Research 86(1), 42–78: doi.org
- Pew Research Center, “How Teens Use and View AI,” February 24, 2026: pewresearch.org
- Education Week, PISA 2025 results, September 2026: edweek.org
Checked October 15, 2026. Spot an error? Tell us and we will correct it.
Frequently asked questions
What is Alpha School?
How much does Alpha School cost?
Does Alpha School really work?
Is there a school with no teachers in Chicago?
Does AI tutoring improve learning?
What should parents ask before choosing an AI school?
Watch: video explainers
A network news report on the new Chicago campus. Compare what you see with the verified and self-reported claims table on this page.
An independent creator's look at the AI school model and its price tag. The research section here adds the peer-reviewed evidence on AI tutoring.
Videos are hosted by YouTube and load only when you press play (privacy-enhanced mode). We are not affiliated with these channels.