Won't bother reading back further, so many replies i totally agree with on this page alone. AI can't analyse and utilise info effortlessly like the human brain can, our brains are around 1.3/1.4 kilos of thought processing dynamite.
We even process information and learn when we're asleep/powered down.
. AI does not need to think better than humans to replace them. There are many jobs that do not require "thinking". You follow directions, do what you are told, and AI can do that better, faster and cheaper. Then if you look at jobs that do require thinking, often the thinking takes up less than half the job.
I was a HS teacher, and my job required thinking. Every year I was trying to improve the results of my class as measured on a statewide exam. I used statistical tools to evaluate how the class was doing on each subject, which approaches had the best result, etc. Perhaps I can do that better than AI can. But that analysis was at most 5% of my time. What takes the most amount of time is grading papers. AI can do that much better than me and I knew it. I had 150 students, I knew that instead of simply telling them if an answer were right or wrong if I analyzed each answer on each paper I could help the kids even more. But I didn't have the time to do that, AI does. I also knew that speedy grading and posting of grades would help parents and students. We once had a student on a Zoom call with the principal complaining about every teacher (except me) not posting grades in a speedy fashion. This brings up another point, not all teachers perform the same way. It could be that AI cannot do the job of 20% of your employees better, but if it can do it better than 80% that would wipe out employment in that industry.
Then of course there is lesson planning. There are decisions made based on my statistical analysis but after that the lesson planning for the most part is following specific directions which AI can do better faster than I can. But here is the thing, every single science teacher in NY teaching my subject is teaching the same curriculum. I can definitely see that if you took the five highest performing teachers in the state and had them create the lessons and do all the statistical analysis on the state to keep modifying the curriculum it would be very hard to argue that the result would not be better than what other teachers could do, especially in the aggregate. It might be that some teacher in some school would have a better outcome for a few students, possible. But when viewed as a whole the state would do much better. But where AI really shines and destroys any other teacher is that it can deliver the lesson individually to each student, monitoring how each student has done. I put a lot of review weaved into my lessons over the course of the year. If that review were taylored to each student, based on what they know and what they don't, it would be far more effective. I can't do that, I can only do it based on what 150 students know and don't know. But AI can. It can know which concepts you are struggling with and give you extra work on those concepts and at the same time identify the concepts you know well, and avoid needless repetition on those concepts.
There you go, I have just described Khan Academy, what they are doing for every single subject. Prior to Khan academy I could understand why parents might fear homeschooling their kids past elementary school. But now I could guarantee that any child who is homeschooled for three or more years will outperform anything they would have done had they gone to public school. Growing up I had four teachers that stood out as extraordinary: 5th grade, 8th grade science, HS English lit, and HS Social studies. In addition to these I would include another four teachers that were very good in my 8th grade but I can't even remember their names. However, sixth grade was a disaster, as was 7th grade, and perhaps 15 of my HS teachers were also forgettable. So out of 8 years you could easily replace 60% of the teachers. So it may be that 40% of a child's teachers will be on the level of Khan academy, but 60% won't.
Something else to realize, the first three years of a teacher's career are below average. This is true of every single teacher, you check and the first three years were their worst years regardless of whether they were OK, good, or terrible. If AI is better and less expensive than hiring a new teacher, then it becomes clear it won't be long before all the teachers are replaced.
I have done other jobs including being a Stock Broker, but nothing demanded more thinking and analysis than my job as a teacher and even I can see that kids would be better off being homeschooled. Now there is value to play on a team, or the band, or to be in a club. But the schooling part can be done much better by Khan academy. Homeschooling does not mean the kid stays at home. On the contrary, you can do it at a church with about 20-30 other kids using the classrooms.
Public school is a disaster. Yes, about 20% of the schools are well run by a very good principal and in those schools they probably have 20% of the teachers who are very good. But even those schools cannot hold a candle to a well designed homeschool run out of a caring church where the kids are also involved in other activities like the choir, band, little league, etc. Since Khan academy is so inexpensive and effective I would say that solves the problems of curriculum design. You still need to have one or two caring people run the program, vetted by the Pastor. But most churches have retired teachers that could assist, and in any homeschool program you ask for parents to assist as well.