Alpha School vs. Traditional School: What Our Family Learned From a Year of AI Learning
The first time I walked into Alpha School for an observation day, I expected something futuristic — robots, holograms, maybe a wall of blinking lights. Instead, it looked a lot like a coworking space for small people. A row of kids sat at desks with headphones on, each working on a laptop. A mentor crouched next to a boy who was confused about a math problem, tracing something on a whiteboard. The girl next to him was quietly recording herself giving a presentation about how to budget for a school fundraiser.
My son, nine, was doing a reading comprehension exercise. The screen showed a passage about beavers and asked him to identify the main idea. When he got it wrong, the program didn’t just mark it incorrect. It gave him a shorter paragraph, then a sentence-level version, until he could answer correctly. The whole thing took about four minutes.
We had enrolled him three months earlier, after a spring of daily worksheets and a teacher who told us he was “bright but easily distracted.” He would come home saying school was boring, then spend an hour on Khan Academy videos just because they let him go at his own pace. We weren’t sure an AI school was the answer, but we were tired of fighting him to finish homework.
Now, watching him work, I saw something I hadn’t seen in a while: he wasn’t trying to finish. He was actually reading.
That first year taught me a few things, and none of them were what I expected.
The biggest surprise was how ordinary the academics were. For a program that promised personalized learning, there wasn’t a lot of flash. Kids spent roughly two hours a day on core subjects using AI software that adapted to their answers. If they mastered a concept quickly, they moved on. If they struggled, the next question got easier, not harder. The software kept a running list of exactly which skills they had and hadn’t internalized.
The rest of the day was split between projects, workshops, and what they called “real-world skill blocks.” One week, my son’s group designed a simple board game and had to pitch it to the other kids. Another week, they planned a mock family vacation with a set budget. The mentors—none of them were called teachers—kept things on track, but they didn’t stand at the front of the room. They worked one-on-one with kids who needed help, while the others worked independently.
The moment I got less skeptical was when my son hit a wall on fractions. He had powered through multiplication and division in a few weeks, but fractions stalled him. The AI would give him a question, he’d get it wrong, the software would show him a hint, and he’d get frustrated. I saw him click through the hints quickly, just to get to the next question, but the program wouldn’t let him progress until he got the concept right.
He finally threw his headphones down. “It doesn’t make sense!”
A mentor named Ms. Ortiz came over and didn’t tell him to calm down. She just sat next to him and asked him to explain what he thought a fraction was. He gave a muddled answer about pieces of a pizza. She drew a rectangle and split it into three columns. She shaded two and asked, “What part is this?” They went back and forth for five maybe ten minutes, until the trouble became clear. The issue wasn’t fractions. He didn’t fully understand what the “top number” and “bottom number” represented. He had memorized the words but not the idea.
The AI had diagnosed his mistakes as a fraction problem. Ms. Ortiz realized it was a vocabulary problem. That’s a distinction software alone probably wouldn’t have caught.
That moment made me realize what “AI school” really means: not machine-led education, but machine-guided education with human judgment on top. The software was great at pacing and practice. It could tell me he was weak in equivalent fractions, but it couldn’t tell me he was imagining a pizza with a giant pepperoni slice labeled “denominator.” The mentor’s role was to see the whole kid, not just the data.
Still, I had a lot of worries at first.
Screen time was the obvious one. When I told other parents where my son went to school, the first question was always, “So he’s just on a computer all day?” That wasn’t true. The academic blocks were screen-heavy, but the project blocks, the outdoor time, and the afternoon apprenticeship periods were hands-on. He was spending maybe two to three hours a day on a laptop, which was actually less time than he’d spent on educational apps at his old school.
Socialization was trickier. There were no bells or hallways, and because every child worked at their own pace, the class wasn’t doing the same thing at the same time. I worried that meant less bonding. But I noticed the kids ate lunch together, played outside together, and worked on group projects constantly. The social dynamics were different, less based on who was good at spelling or who could sit still. It was more collaborative, but it wasn’t magic. Kids still argued. My son still had days he didn’t want to go.
What I wasn’t prepared for was how much my son would start talking about his “level.” The software displayed progress bars and percentage completions, and he loved watching his numbers go up. But it also made him anxious when he wasn’t advancing as fast as a friend. We had to talk a lot about what the numbers actually mean. I don’t think that’s unique to Alpha School. Any system that gives kids data about themselves will create some stress. But it did mean more conversations at dinner about effort versus speed.
If you’re considering an AI school for your child, I’d suggest asking some practical questions before enrolling. First, ask to observe a full morning. Not a tour. A real observation where you sit in the room and watch for an hour. Notice what the mentors are doing. Are they circulating, asking questions, and checking in with kids, or are they mostly managing the software? The quality of the human adults matters more than the quality of the software.
Second, ask what happens when the AI doesn’t work. Every adaptive learning program has gaps. Math concepts that require drawing or manipulating objects, long writing assignments, and open-ended science questions often still need a person. The school should have a clear answer for how mentors step in.
Third, ask about transparency. Can you see your child’s actual progress notes? At Alpha School, parents get updates, but they’re not always as detailed as I wanted. I had to ask directly for samples of my son’s writing and project work, not just his score reports. A good school won’t hide those things, but you might have to request them.
I also think it’s healthy to ask what the school isn’t good at. Every school has weaknesses. Alpha School is not great at traditional group instruction. If your child thrives on listening to a teacher explain a topic with lots of discussion and debate, an AI model might feel isolating. My son is quiet and likes to work at his own pace, so it fit him well. His friend who loved class discussions didn’t adjust as well.
By the end of the year, I still had mixed feelings. My son’s reading had improved more than it had in two years at his old school. He could explain his math reasoning better than most adults I know. But he was also more comfortable sitting alone with a screen than I was entirely happy about. He sometimes missed the structure of a regular classroom, even while he appreciated the freedom.
We’re still at Alpha this year, but we’re reviewing it each term. I’ve stopped thinking of it as an “AI school” or a “normal school.” It’s just school. It has pros and cons, like everything else. The AI handles the pacing and the practice. The mentors handle the hard part — seeing the child underneath the data. That division works, most days. I just don’t know yet if it’s enough for the long haul. But for now, it’s working, and that’s more than I could say a year ago.
Please indicate: Thinking In Educating » Alpha School vs