The $55K AI School With No Teachers — And Why It Misses the Point

A $55,000 Experiment in Replacing Teachers With AI

This fall, a private school called Alpha Schools will open its doors in Chicago’s Lakeshore East neighborhood. There will be no classroom teachers. No lectures. No lesson plans delivered by human educators. Instead, children in pre-K through eighth grade will spend two hours each morning learning from AI-powered software while adult “guides” — who don’t need education degrees — focus on motivation and emotional support.¹

The price tag? Fifty-five thousand dollars per year.

Now, you might be thinking: “If AI can teach, why do we need teachers at all?” That question is exactly why this matters — and exactly where the thinking breaks down.

The Alpha Model Fallacy

Here’s what Alpha Schools gets right: the traditional model of one teacher lecturing thirty students at the same pace is not optimal. Decades of research confirm this. A 2025 randomized controlled trial from Harvard found that AI tutoring outperformed in-class active learning, with students learning significantly more in less time.² That’s a real finding from a real study. AI-powered personalization works.

But here’s what Alpha Schools gets catastrophically wrong: they’ve confused the tool for the teacher.

Alpha’s model compresses core academics into a two-hour AI block, then fills the rest of the day with life-skills workshops, entrepreneurship activities, and project-based learning. The company claims their students rank in the top 1% nationally on standardized tests and grow at 2.6 times the rate of peers on MAP assessments.¹

Those numbers sound impressive. They’re also unverifiable by independent researchers, drawn from a tiny sample of self-selected wealthy families, and — critically — they measure the exact kind of narrow academic output that AI is optimized to produce. Standardized test scores are AI’s home court. That’s like measuring a calculator’s success by how fast it does arithmetic.

Charles Logan, an education researcher at Northwestern University and former public school teacher, put it directly: the research on personalized AI learning used as a primary method of instruction is “mixed at best.”³ A 2025 systematic review of 28 studies involving over 4,500 K-12 students confirmed that while AI tutoring systems show generally positive effects, those advantages shrink significantly when compared to any form of structured human tutoring.⁴

The Real Crisis Isn’t Teaching — It’s Training

While Alpha Schools builds a boutique escape hatch for families who can afford $55,000 per child, the actual AI literacy emergency is happening in every public school district in the country. And it’s not about replacing teachers. It’s about the fact that we’re not equipping them fast enough.

RAND’s nationally representative survey data from 2025 reveals a staggering disconnect: 54% of students now use AI for school-related tasks — up more than 15 percentage points from just a year earlier. Yet only 35% of school districts provide students with any AI training at all. Over 80% of students reported that their teachers have never taught them how to use AI for schoolwork.⁵

Read that again. The majority of students are already using AI. The vast majority have received zero guidance on how to use it well.

This isn’t a technology problem. It’s a teacher empowerment problem. When educators get proper AI training, they don’t become obsolete — they become more effective. They use AI to identify struggling students faster, differentiate instruction more precisely, and spend less time on administrative tasks that drain their energy.

The Equity Problem Nobody’s Talking About

Alpha Schools’ founder, Mackenzie Price, told CBS Chicago that the school’s model allows AI to “raise human intelligence.” But there’s an inconvenient math problem with that vision: at $55,000 per student, this version of “raised intelligence” is reserved for roughly the top 2% of American families by income.¹

As of this writing, the Chicago campus has two enrolled students. Thirty-five have expressed interest. The school aims for 50 by fall.

Meanwhile, 54 million children attend public schools where the question isn’t whether AI will transform education — it already has — but whether anyone will teach them how to navigate that transformation responsibly. Only 45% of school principals report having any AI policy in place. Just 34% of teachers say their schools have policies specifically addressing AI and academic integrity.⁵

The Alpha model doesn’t solve the AI literacy crisis. It sidesteps it entirely — and only for those who can write a check.

What the Research Actually Tells Us

Here’s the part that gets lost in the “AI replaces teachers” headlines: the most compelling research on AI in education doesn’t support replacement. It supports integration.

The Harvard study that Alpha Schools’ defenders love to cite? It compared AI tutoring to active learning — not to teachers working alongside AI tools. The researchers themselves note the study’s limitations: a two-week intervention, college-level participants (not elementary students), and no measurement of the social-emotional development that happens through human teaching relationships.²

Khan Academy’s Khanmigo platform grew from 40,000 to 700,000 students in a single year — but Khan Academy has always positioned its AI as a supplement to teachers, not a substitute.⁶ That’s not a small distinction. That’s the entire distinction.

Liz Gerber, a researcher at Northwestern’s Center for Human-Computer Interaction, pointed out something that should give every parent pause: Alpha’s approach isn’t actually new. “It’s personalized learning,” she said — a concept that has been around for decades, with results that have consistently underperformed the hype.¹

The SeedStacking Alternative

At Harvest Kernel, we’ve built our entire methodology around the opposite premise from Alpha Schools. SeedStacking doesn’t replace human capability with AI — it uses AI to amplify what humans can already do, one small daily practice at a time.

The difference is philosophical, but it has real consequences:

The mistake most people make — and the mistake Alpha Schools is making at scale — is treating AI literacy as a product you can buy. Install the right software, hire some “guides,” and watch the test scores climb. But AI literacy isn’t about AI doing the work. It’s about humans understanding AI well enough to use it wisely, critically, and confidently.

That doesn’t come from two hours of adaptive software. It comes from what happens when a teacher sees a student struggle with an AI tool, recognizes the confusion, and turns that moment into a lesson about critical thinking. No algorithm has that instinct. Not yet. Maybe not ever.

The Question We Should Actually Be Asking

Alpha Schools’ expansion to Chicago will generate headlines. It will spark debate. Some parents with means will enroll their children. And the education technology industry will hold it up as either a warning or a triumph, depending on which narrative sells better.

But the question worth asking isn’t whether AI can teach children. The question is: what kind of learners do we want to build?

If the answer is “students who score in the top 1% on standardized tests,” then by all means — hand them a laptop and an algorithm. But if the answer is learners who think critically, collaborate authentically, navigate ambiguity, and use AI as a tool rather than a crutch — then you need teachers. Real ones. Empowered, trained, AI-literate educators who understand both the capability and the limitations of the technology that will define their students’ futures.

That’s not the expensive path. It’s the scalable one.

Start small. Build real capability. Let it compound. That’s what SeedStacking is built for — and it works whether you have $55,000 or $0 to spend.

Sources

1. CBS News Chicago, “Alpha Schools, which uses AI instead of teachers for learning, is enrolling in Chicago for fall 2026,” March 2026
2. Kestin et al., “AI tutoring outperforms in-class active learning,” Scientific Reports, June 2025
3. Block Club Chicago, “An AI School, With No Teachers, To Open in Chicago This Fall,” March 2026
4. npj Science of Learning, “A systematic review of AI-driven intelligent tutoring systems in K-12 education,” May 2025
5. RAND Corporation, “AI Use in Schools Is Quickly Increasing but Guidance Lags Behind,” 2025
6. K-12 Dive, “3 questions for K-12 leaders to consider amid the AI tutoring boom,” August 2025