Schooling vs. Learning
Today is September 4, 2026. GPT-6 Astra came out today (though access is still not rolled out), and we saw Anthropic release Fable 5.1, Google release Gemini 3.8 Flash, and Facebook release Muse Spark 1.3. This is all while local AI is getting crazy exciting. It truly feels like a renaissance moment, objectively speaking, and things are only gonna accelerate further. AI is penetrated very little relatively, and there is so much capital and energy pointed at it.
Contrast this with some headlines getting shared in the broader/political-sphere: Zohran bans Generative AI use for some public school students, and Sanders proposing legislation to “Ban Artificial Superintelligence”.
There’s so much to unpack, but if you try to unpack it from anywhere except from the bottom rung of the argument, the foundation, you will get spun out. Take something like this:
Person 1: “Zohran banned AI”
Person 2: “Good, the people pushing AI are the same people pushing us into a feudal state, they are techno-facists eroding democracy”.
Or, take this:
Person 1: “Zohran banned AI”
Person 2: “Have you seen the video of the high school teacher in Houston crying because his students can’t read?”
So, see the problem – there are so many levels to come at a topic like this, because it’s now politicized, it involves a lot of money, power, and it’s all underpinned on fundamental (mis)understandings of how technology works and what the definition of AI is. And that’s all to be expected. I see no way that AI isn’t one of the major political topics from 2027 and beyond.
Instead of going down these sorts of rabbit holes and measuring who wastes the most water and electricity between Data centers, golf courses, almond farms, and Carnival cruises, I want to ponder on a clearly framed fundamental question, and conjugate it until we feel comfortable.
Let’s start here:
“Assuming AI can perform a cognitive task for us, what do we still need to learn ourselves?”
The first thing that jumps out at me is, “what’s the purpose of learning?”. It’s implicitly assumed when talking about this stuff. But there are very different answers. First problem is, what does learning mean? Learning, in what sense? Do you mean learning in a way that is interchangeable with “schooling”? Or do you mean capital “L” learning, like, some sort of platonic form Learning, an idealized and more personal concept of learning?
Let’s take the second one because it’s easier. If you mean your own personal learning, there’s not much debate to have, there is only interesting discussion and conversation. It would be drastically different for everyone, because everyone’s brains are so different and everyone learns differently. The answer would basically be, “we should use it to the extent that our pareto-optimal learning level is achieved”. Kind of a non-answer, but also, it makes sense. We would need to use AI insofar as it helps us achieve our task of learning. To figure that out would involve curiosity and experimenting, setting up systems, tracking, and updating priors, and getting good at using the technology itself, becoming accustomed with it, etc.
Okay, so that’s the second route. Let’s talk about the first one. Because this is the one that people usually mean when they say “learning”. In this one, they mean, more precisely, “schooling”. And a lot is packed into this. They are basically opining about how people should be schooled, in general. So, in order to have an opinion about this, we need to ask another question:
“What’s the purpose of schooling?”
Even this is loaded with assuming “public schooling”, let’s just assume this is public schools, not taking into account private schools, homeschooling, Montessori, etc. All of these would claim different answers. But since the majority of people receive public schooling, let’s stick to that. So, what’s the purpose of (public) schooling? Is it to facilitate the process of personal knowledge attainment ? E.g., the same goal as the second thing, but just applied across N number of students?
I would argue that’s a resounding “No”, it’s kind of for something different. It’s more to get you ready to be a participating member of society. You should be able to get places on time, turn in assignments, read, write, do math, know historical events, scientific facts, and they learn some socialization, that sort of thing. The modern American schooling system was basically created by Horace Mann in the early 1800s, influenced heavily by Prussian schooling systems. By 1970, writers like Ivan Illich, John Holt, and Alvin Toffler had begun to reflect on this question of Schooling Vs. Learning in the modern age. Toffler, for example, popularized the idea that the modern schooling system’s fundamental purpose is to produce people that are adapted to industrial civilization. Illich talked about the fundamental differences between learning and schooling. And Holt talked about the ways that compulsory schooling can negatively affect the way children naturally learn and grow.
My point here isn’t to make it seem like schooling is bad or be a conspiracy theorist or communist, it’s more to say that, the real purpose of schooling in reality is much closer to an institution that generates people who are ready to enter the workforce, pay taxes, and participate in the economy, than some sort of personal knowledge attainment and a realization of personal ideals, goals, and knowledge. It’s an appendage of the state.
So, what’s the answer to our original question?
“Assuming AI can perform a cognitive task for us, what do we still need to learn ourselves?”
The answer, is, “it depends on what you mean by learning”.
If we want to be more precise and ask something like:
“Assuming AI can perform a cognitive task for a child in school, how do we decide when to allow them to use AI and when they can’t?”
I also think it still depends. Even the most AI-pilled people will agree that some baseline of knowledge is required in order to interface with AI. But where do we draw the line? Again, it’s complicated.
One thing I want anyone making laws about this to understand is working with AI is an essential skill in the future workforce, so if you don’t allow your students to use it, you will be hindering them. But there is some appropriate level of constraining the use of it.
For example, I remember being able to use materials that would help me for homework and some quizzes, but not during the tests, in some classes. This needs to be up to the individual educators. But if AI is this politicized, polarized thing, teachers are gonna have their bias as well, and it will probably be to the anti-AI side, and this will hurt students.
Can write a lot more about this, but I hope people on both sides of the AI debate will think critically and find common ground and common sense.