When AI Replaces the Teacher: Why Connection Still Matters More Than Efficiency

By Michelle Collins, Chief Revenue Officer, CodeBaby

I was reading the New York Times piece about Alpha School and I couldn’t shake this feeling that we’re witnessing a fundamental misunderstanding about what education actually is.

Here’s the setup: Alpha School in Austin uses AI to handle all core academic instruction in just two hours a day. Students work with apps and algorithms to learn math, reading, and science while adult “guides” provide motivation and emotional support. The rest of the day is spent on practical skills and passion projects. MacKenzie Price, the founder, calls it the future of education and is expanding to more than a dozen cities this fall.

The results sound impressive. Students reportedly learn “twice as fast” and rank in the “top 1-2% nationally.” Eleven of their first twelve graduates went to four-year universities. The efficiency gains are real.

But here’s what keeps nagging at me: when did we decide that education is primarily a content delivery problem to be optimized?

Because that’s essentially what Alpha School is proposing. Strip away the inefficiencies of human instruction, let AI handle the information transfer, and use humans only for the “soft skills” stuff that algorithms can’t manage yet.

It’s a seductive pitch, especially when traditional classrooms are struggling with overcrowding, teacher shortages, and one-size-fits-all approaches that leave too many kids behind. AI can absolutely deliver personalized instruction at scale in ways human teachers simply can’t match.

But it misses something fundamental about how learning actually happens.

The Curation Problem

Let’s be honest about what Alpha School actually is: it’s a $40,000-per-year private school that admits selective students and employs guides making six-figure salaries at 5:1 adult-to-student ratios. When Pennsylvania’s Department of Education rejected their charter application, they noted that the model was “untested and fails to demonstrate” alignment with academic standards.

That’s not a bug in their system – it’s a feature. Alpha can show impressive results because they’ve created optimal conditions that have very little to do with AI. They’ve curated their student body, hired exceptional staff, and created resource-rich environments that would make any educational approach look successful.

The problem comes when this boutique model gets held up as proof that AI should replace teachers everywhere. As Randi Weingarten from the American Federation of Teachers pointed out, “Students and our country need to be in relationship with other human beings. When you have a school that is strictly A.I., it is violating that core precept of the human endeavor and of education.”

But here’s what really concerns me: the narrative that this represents some inevitable future where human teachers are obsolete. Price calls classrooms “the next global battlefield” and writes on social media, “I’ve seen the future, and it isn’t 10 years away. It’s here, right now.”

That kind of rhetoric troubles me because it fundamentally misunderstands what education is actually for.

What We’re Actually Optimizing For

When Justin Reich from MIT’s Teaching Systems Lab says, “If you think of the purpose of schools as to prepare people for the roles of citizenship and democracy, there’s lot of places where you aren’t trying to get kids to race as fast as they can,” he’s pointing to something crucial that gets lost in the efficiency narrative.

Education isn’t just about content mastery. It’s about learning to think critically, to collaborate with people who disagree with you, to navigate complex social dynamics, to persist through challenges, and to develop the kind of judgment that can’t be algorithmically derived.

Alpha students do work on collaborative projects – building food trucks, creating mountain bike parks, developing chatbots. But when the core intellectual development happens in isolation with an AI tutor, you’re missing the messy, irreplaceable process of learning alongside other humans.

The article mentions that some Alpha students leave after middle school to experience “team sports, student council and prom night” at traditional high schools. That tells us something important: even students thriving in this system recognize there are aspects of human development that require genuine community and shared struggle.

And here’s what really gets me: new research is showing that heavy AI reliance in learning actually weakens neural connectivity and memory formation. When students become dependent on AI for thinking and problem-solving, they’re not just learning differently – they’re potentially losing cognitive capabilities they’ll need throughout their lives.

A Different Kind of AI in Education

I’m not anti-AI in education – far from it. But I think we need to be a lot more thoughtful about how we deploy it.

The Alpha model assumes that the best way to use AI is to have it replace the core function of teaching. But what if that’s backwards? What if the most powerful application of AI in education is to enhance rather than replace human connection?

At CodeBaby, we’re working on something fundamentally different. Instead of AI tutors that deliver instruction, we’re developing study buddies that help students prepare for what they’ll do in the classroom. Our avatars don’t give students the answers – they pose questions, guide them when their responses aren’t quite right, and help them work through problems so they arrive at class better prepared to engage with their teachers and peers.

It’s the difference between replacing the learning relationship and strengthening it. When a student comes to class having already wrestled with concepts through guided questioning, they’re ready for deeper discussion, more complex applications, and the kind of collaborative problem-solving that can only happen between humans.

This approach recognizes something that the Alpha model seems to miss: the real learning often happens not when you get the right answer, but when you struggle with the question. When you have to articulate your thinking, defend your reasoning, or reconsider your assumptions based on feedback from someone who sees things differently.

Our study assistants are designed to prepare students for that human interaction, not replace it. They help build confidence and foundational understanding so that classroom time can be spent on the higher-order thinking that actually transforms how students see the world.

The Human Elements That Still Matter

Alex Mathew, a 16-year-old Alpha student, said something telling in the Times piece: “To be a useful person in the age of A.I., you have to have unique insights that A.I. doesn’t really agree with. That’s the real differentiator. We are trying to beat A.I.”

I appreciate that perspective, but it reveals something troubling about how we’re framing the relationship between humans and AI. If the goal of education becomes “beating AI” or developing “spiky points of view” that algorithms can’t replicate, we’re setting ourselves up for a pretty bleak future.

Because the most important human capabilities aren’t about being more creative or contrarian than machines. They’re about being empathetic, collaborative, ethical, and wise. They’re about developing the judgment to know when to trust technology and when to question it. They’re about learning to work with people who are different from you, to navigate ambiguity, and to persist through challenges that don’t have clear solutions.

Those capabilities develop through relationship and community, not through individual optimization. They emerge when students learn to see themselves as part of something larger than their own academic achievement.

When I think about the most transformative educational experiences in my own life, they weren’t about information transfer. They were about connection. The classmate who saw potential I didn’t know I had. The mentor who challenged me in ways that made me think differently about myself. The educator who created an environment where it felt safe to be curious, to ask questions, to fail and try again.

You can’t get that from an app. And you can’t scale it through “motivational guides” operating at arms-length from actual instruction.

Building Technology That Enhances Humanity

The question isn’t whether AI belongs in education. It’s how we use it in ways that strengthen rather than replace the human elements that make learning meaningful.

This distinction matters more than we might realize. When Alpha students spend their core academic time interacting with algorithms rather than wrestling with ideas alongside teachers and peers, they’re missing something essential about how knowledge actually develops and deepens.

Our study buddy approach at CodeBaby is designed around a different philosophy entirely. We want students to arrive at class having already engaged with material through guided questioning and supportive feedback. When they sit down with their teacher, they’re not hearing concepts for the first time – they’re ready to explore implications, make connections, and tackle the kind of complex applications that require human insight and collaboration.

The goal isn’t to make classroom time obsolete. It’s to make it more valuable. When students come prepared, teachers can focus on the higher-order thinking, the nuanced discussions, and the collaborative problem-solving that actually develops critical thinking skills.

But they’re tools that work alongside human educators, not replacements for them. Because the goal isn’t to prove that technology can do everything humans can do. It’s to create systems where technology handles what it does best so humans can focus on what they do best.

And here’s the crucial difference: our avatars are designed to foster genuine connection while providing the efficiency benefits that institutions need. They remember what students have told them, adapt to different learning styles, and provide the kind of patient, consistent support that helps learners build confidence. But they’re explicitly designed to prepare students for meaningful human interaction, not replace it.

The Real Test

Pennsylvania’s Department of Education rejected Alpha School’s charter application, noting that “the artificial intelligence instructional model being proposed by this school is untested and fails to demonstrate how the tools, methods and providers would ensure alignment” to educational standards.

That rejection points to something important. Before we scale these models, before we hold them up as the future of education, we need rigorous, independent research on what they actually accomplish and what trade-offs they require.

We need to understand not just whether students can achieve higher test scores, but whether they’re developing the critical thinking, social skills, and resilience they’ll need as adults. We need to know whether this approach works for students from different backgrounds, with different learning styles, facing different challenges.

And we need to be honest about what we’re optimizing for. If the goal is to create more efficient content delivery, then yes, AI tutors probably win. But if the goal is to develop confident, capable, empathetic humans who can think critically and work collaboratively, then we need to be a lot more careful about what we’re automating away.

Because at the end of the day, education isn’t just about what students learn. It’s about who they become. And that transformation happens in relationship with other humans who see their potential, challenge their assumptions, and support them through the messy, non-linear process of growing up.

No algorithm can replace that. And honestly, I don’t think we should want it to.