When I was young I lived on science fiction from authors like Isaac Asimov that regularly featured tons of technology including robotic teachers and technology that had general artificial intelligence. That lead to an interest in computers and in 1995 I enrolled in a computer science program at a state college and I currently work with data collections that come from student information systems. I also have a kid in high school right now.

My dad worked at a university and we had an electric motorcycle and car prototypes in the late 80's/early 90s that used racks of car batteries to drive their electric motors. Man that bike was bulky, but it was quiet and the potential was obvious once the battery bulk was addressed. It has taken decades to get there, but now we have tons of electric vehicles and that is awesome! Hell, I didn't expect the miniaturization of computers to fit in the current form factor in my lifetime.

But the thing is, having the idea and the possibility of something involves multiple parts. Electric vehicles wouldn't be possible at scale until they solved the battery bulk problem. Working towards it was good, as there was an end goal that could be met if only we could solve the form factor. The whole thing with AI not being a realistic thing for replacing teachers is that there is far more to it than a computer and it will require hardware advancements that are pie in the sky at the moment in addition to the actual process of machine learning. Science fiction modeled advanced computers after the neural networks of the brain, such as Asimov's positronic brains. For AI to be at least as smart as the average person they will need a network of a similar complexity. We are currently using massive amounts of energy to regurgitate jokes from reddit as facts. To have something like a teacher we will need to make the energy requirements similar to a brain, in a form factor similar to a brain, and with the ability to learn and share with other fake brains to improve over time.

But then we get to the reality part which is nobody is going to do this work at scale to make pretend teachers of any quality any time soon. Student software systems used by schools are pretty rough because the money isn't there. That is just tools for people to use, not as a replacement for people! There is far more to being a teacher than the subjects they teach. They need to motivate students, act as human beings in real meat space, and even then a lot of teachers are mediocre but trying and some are even terrible because teaching is hard.

Now you might say that ai home schooling would be done by a motivated kid, but a motivated kid doesn't need ai to learn on their own. They do that already! We already have tons of available content like Khan Academy that motivated kids can use to learn. But Khan Academy can't motivate a child to learn, and ai won't either because it is a thing and not a person who has life experience and a complex biology in addition to their brain that can improve success in teaching. For ai to be used to teach more than 50% of students it would need to be at least as good as the better teachers including the ability to motivate the students.

Between my wide eyed youth and today I have seen a ton of technology that gets touted as the next big thing and some are some are hits and some are obvious misses. There are tons of technologies that are promoted for specific uses that are clearly bullshit either due to the technology itself or because it is being sold as a magical cure all by snake oil salesmen. AI as a magical cure all is snake oil. It has a lot of uses, but this is not one of them.

Extra credit: Home schooling is fine for some kids, but the vast majority of kids need to share physical space with other children outside their home as part of their development because we are biological creatures. Even in 1000 years the majority of students learning exclusively alone at home from any kind of teachers is a pipe dream.