One Parent Wants to Move the Kids to an AI School. Now What?

School Choice Is Almost Always a Joint Decision

Most custody orders separate physical custody, which governs where a child sleeps, from legal custody, which governs major decisions — medical care, religious upbringing and education. Joint legal custody is the common arrangement, and under it neither parent can unilaterally change a child’s school. Enrolling a child somewhere new without the other parent’s agreement is one of the faster ways to end up back in front of a judge, even when the enrolling parent genuinely believes the move is better for the child.

So the order of operations matters. Read what your order actually says about educational decisions before you tour anything, submit an application or put down a deposit. If the order is silent or ambiguous, that ambiguity is not permission — it is the thing to resolve, in writing, before money changes hands.

What the Evidence Supports, and What It Does Not

It is worth separating the school from the method, because they are not the same claim. Alpha School, the school in most of these videos, charges between $10,000 and $75,000 a year depending on the campus, with roughly $40,000 the common figure across its thirteen campuses. Its results come from its own internal testing analyses. No peer-reviewed study has evaluated the model, nobody independent has audited the data, and Pennsylvania’s education department called the approach untested when it denied a related application in 2025.

The method underneath it has far better support. A Harvard physics trial published in Scientific Reports in June 2025 had 194 students each learn one topic with an AI tutor and another in a well-run active-learning class. Median gains in the tutored condition were more than double, and students finished in a median of 49 minutes against a 60-minute class. A World Bank trial in Nigeria found a smaller but real effect after six weeks of after-school AI tutoring.

Notice what that result actually is. The advantage showed up as compression — the same learning in less time — which is the entire two-hour school day. It is also fair to say what the trials did not measure: retention over months, transfer to harder problems, or what a year of this does to a nine-year-old. Our studio’s evidence teardown of the two-hour school day goes through which claims survive scrutiny and which do not.

The Part That Costs Nothing and Works in Both Houses

Here is the useful news for families who will never write a $40,000 check, and especially for families running two budgets. Three habits carry most of the effect, and none of them require a campus.

Pace by mastery rather than by the calendar: a topic the child has not cleared keeps coming back, and a topic they own never appears again. Test to learn rather than to measure, because retrieving an answer builds the memory and re-reading the chapter does not. And close the feedback loop the same day — a practice test graded a week later teaches almost nothing, because the child no longer remembers what they were thinking.

Those three behaviors are what ExamPilot is built to run, and the reason they matter here is continuity. A child who studies one way at one parent’s house and a different way at the other’s loses most of the benefit. A shared progress record means the parent who has the child on Thursday can see exactly what Tuesday’s session covered, without either parent having to ask.

One more thing worth knowing before you sign anything: when a child’s whole education runs through software, somebody owns the record it produces, and it is usually not the family. Our sister site Nexaria Digital looks at who holds the transcript when the classroom becomes an app — a question worth putting to an admissions office directly.

Put the Decision in Writing Before the Deposit

Whatever the two of you decide, write it down while you still agree. Name the school, the start date, who pays what share of tuition and fees, how a mid-year withdrawal would be handled, and who holds the login credentials and the academic records. Agree on what happens if one parent wants to reverse the decision later.

It is also fair to say out loud that the current baseline is not sacred. Federal NAEP results for 2024 put eighth-grade public-school math 8.8 points below where it stood in 2019 and flat since 2022, with reading down in every cycle and about a third of eighth graders reading below the basic level. Wanting something different for your child is a reasonable instinct. The obligation is to make the change together, with the evidence in front of both of you rather than a video.