Alpha School's 'AI Instead of Teachers' Model Is Being Overhyped
In this video, Dr. Justin Baeder discusses why the Alpha School model of replacing teachers with AI is getting more hype than it deserves.
Key Takeaways
- The model replaces teachers with screens - Alpha School uses AI-driven instruction instead of certified teachers
- Hype outpaces evidence - There's insufficient evidence that AI-based instruction works better than teacher-led classrooms at scale
- Human instruction matters - The relationship between teacher and student is a critical component of learning that AI cannot replicate
Transcript
Tuition is $40,000 a year and there are no teachers.
This is Alpha School, the private school chain that was founded in Austin and is now expanding nationwide that promises to help students learn much faster by putting them on computers with AI technology for two hours a day and then having them do non-academics in the afternoon and have social interaction and group projects and things like that.
I think the best case to be made for this model is in math, where a lot of students want to work at their own pace.
You need a lot of practice, a lot of differentiation.
There's a lot of opportunity to help students understand and move quickly in math.
I have to ask, though, isn't this kind of a grim school day?
And is there really much upside in terms of learning faster in other subjects?
I'm not sure that there is.
Like if you are in a social studies class, can you learn much faster with AI?
Is this the equivalent of basically just putting the video on 2x speed or 3x speed?
Like what do we gain by trying to accelerate things that are supposed to be rich, that are supposed to be thought-provoking, that are supposed to be you know, experiences that stick with kids.
Like, I don't know how you do that well individually on a computer with AI.
I think there are a lot of experiences that produce learning that have to be social, right?
Like a class discussion, a debate, you know, maybe we would let go of group projects.
You know, if you did all the heavy lifting on a group project, you might say, well, I could, I could stand to work by myself on some things.
I'm not mad that Alpha School exists, but I think we need to really temper our expectations for how much more learning is going to occur under these models, because I think the biggest constraint on learning is not explanations.
A lot of people want to replace human teachers with AI that can explain things better and faster and not make mistakes and things like that.
Okay, well, eventually a kid you stick on a computer is going to run out of steam.
And when that happens at Alpha, they have all of these extrinsic motivators and rewards and points and requirements and like they won't give you your snacks maybe until you finish your lessons.
And that just seems kind of grim to me like I think our main The motivators should be the content and the, like, the participating in the class itself.
That should be the main sources of motivation.
And that seems to work really well.
Like, most kids are motivated to be in class with their peers.
They're upset if they don't get to be in class with their peers.
You know, and almost...
Alpha school seems kind of like punishment, the kind of thing we would say, you know, you're an in-school suspension, you have to go learn on a computer.
So I think this appeals to a certain kind of tech person.
I don't think it's going to dramatically change the education profession, and I don't think it's going to result in much more learning.
But let me know what you think.