Jared Cooney Horvath on AI schools

Is this the gratification-addiction-zombification of America’s children? Are we about to do a Walking Dead with 4th graders?

By “classroom technology” he’s talking about… what exactly?

The most recently I was in public school was pre-pandemic. We received school-issued Macbooks, and before that we had iPads. I remember us playing games on them twice as often as producing actual schoolwork, despite Appstore restrictions. (Ah yes, Jetpack Joyride…) But we did use them for actual reading, writing (word processing software), and content management (Schoology, Edmodo). Granted, this was before AI took over. These are the devices Horvath is referring to.

What difference does it make if you’re reading in a book or on a screen, anyway?

There are some minor, noticeable differences. Reading a dense article on paper gives you room to slow down and notice overall structure, but a screen has some display differences that influence reading psychology: multitasking, switching between screens, skimming sections, or off-task temptation.

By itself, the claim that “screens cause smooth brain” is weak. Boy… by that logic, all college students must be developing smoothbrain because in my class in chemical engineering, rarely any of us carried around physical textbooks; All our work was done on screens or in spiral notebooks (though I did have a few friends that liked to print things out). There may be an emerging problem of fragmented reading habits even at the college level.

To say “Screens can encourage shallower reading” is more defensible. Screens and electronics don’t slow learning on their own. For short assignments and discussion posts, screens are probably fine. For dense or heavy technical theory, physical print is recommended to help you internalize and retain more information.

Here’s the source video, by the way:

Horvath’s arguments in his most recent book “The Digital Delusion” don’t perfectly align with the broader research landscape in this area, but his overall claims are strong and well-supported. Plainly, “more tech” doesn’t assume “better learning”. Second, the modest improvements device-heavy schooling offers may not justify the cost to implement them. A better idea is to use technology more selectively, which Horvath advocates, rather than to radically cut it out of schools entirely.

What is not explicitly true: “Gen Z is the first generation in modern history to under-perform their parents”–This is highly overstated, and not all of it is because of electronics. There have been systemic changes in K-12 schools unrelated to educational technology (EdTech) that are causing performance to decline, as I pointed out in my last post.

My gut tells me Horvath is pointing in the right direction. He’s exposing a lot of bullshit around the swollen market value of EdTech, but the statistics he mentions are a bit rocky and overblown to sell his point.

I think people’s main fear is the teacher-burnout version of EdTech–That’s where software begins to replace instruction rather than support it. Districts that are already stretched thin may be tempted to use EdTech inappropriately. In those cases it becomes a “crutch”, like watching videos instead of in-person teaching and auto-graded systems instead of human feedback. The distance EdTech creates between an instructor and the student can inadvertently encourage teachers to fall asleep in the control room rather than guide students directly.

AI Private Schools / Alpha School

From what I hear, Ted Cruz has been a proponent of integrating AI into schools in Texas, and you can see him giving Horvath the cowboy stare at the beginning of the video. From what it seems, this may be a disagreement with what Horvath was arguing.

Texas schools have been pioneering the AI-powered education trend, most notably with the rapid expansion of Alpha Schools. AI schools? What are kids doing all day, sitting in front of a computer? Is there an AI hologram of a teacher?

Kids with snotty fingers pointing at laptops with sussy amongus stickers.

Don’t get too fired up about AI teaching your kid instead of a real adult. Alpha School is basically Rushmore with an AI label for public attention and investment. Is it effective? For $40,000 a year, you better believe it. The school’s financial guide says the tuition is meant to cover things like materials, technology, guest speakers, and other fancy stuff (things that would otherwise be billed separately at other private schools).

Each morning they sit in front of a computer and do “AI-personalized academics” for 2 hours, then in the afternoon they work on projects accompanied by “guides”–Humans to provide group interaction and coaching during workshops. The workshops teach practical skills like public speaking, entrepreneurship, coding, arts, and sports.

What sets them apart is the 2-hour learning block with the computer and AI-personalized pacing. Traditional schools focus on repeat instruction and coursework, while Alpha consolidates that and optimizes it at maximum efficiency with their AI thing, then the rest of the day is available for practical, hands-on learning. So it’s a private school that uses AI to optimize their curriculum and track student progress. Can’t public schools just do that?

The answer is yes, actually, many K-12 schools are already. The idea of “AI-personalized learning” is not unique to Alpha Schools. But public schools have more accountability concerns; They have to prove AI integration is making improvements in order to justify the cost of using it. Public schools also have to figure out how to fit AI in their current administrative model, and then there are the typical roadblocks: privacy, training teachers to use it, and district approval. Alpha Schools are selling because they package all that in one price. It’s less about the AI itself and more about their small-campus learning environment.

So… if Alpha Schools are using AI technology selectively, is that a healthy approach to schooling? It’s just 2 hours of screen time a day, tailored by AI, and there’s still interpersonal collaboration and everything. What could possibly go wrong? What would Jared Horvath have to say about that?

As it turns out, he has publicly commented on Alpha Schools. …Wiiiiith Vsauce music.

“That school costs $45,000 a year to join, and they vet their students…” with selective entrance exams and then use their testing results to market the effectiveness of their own program. Guess what, Doctor Harvard? That’s virtually every prestigious learning institution in the United States. They could be motivated by profit or by personal glory–both are arguably predatory. Too bad he spent more time discrediting Alpha Schools than explaining why the model itself is ineffective. He may also be insulting all the students enrolled there. Collateral damage.

Some AI use in schools can be acceptable depending on what it’s optimizing, but there’s no harm in holding off on its deployment for now, at least until we know more about its effects on learning. The kids might have to be experts in it eventually.

BUH–