AI is the teacher now. It's how people are going to learn how to program from now on. They're going to be generating their own examples and have it explain the code to them.
Garry, sorry to say, but you are wrong.
You'd think that is the case (because you know how to program already), but the truth is that anyone below average abuses it to get through rather than benefit from AI.
I have seen graduates who can't write a SLOC in Python because they cheated all the way through CompSci UG with AI. I was forced to work with them.
"Working" with such "people" is a chore that I wouldn't wish upon anyone.
They're essentially credit stealers on steroids, what AI unfortunately is.
The problem is that AI/LLMs aren't a good knowledge extraction tool. Actually, the opposite, you NEED to correct it like a babysitter tries to instill any added value to a newborn. It will give you ideas that are outright crap, and if and ONLY if you are smart enough to notice whenever it is bullshitting you, you will benefit from it. If Vacuum tube computers were crap machines, Zuse would probably call anything LLMs "industrial human waste generators".
Other than that it's just a complete regression.
Literally the thesis of data mining is that the models should AID us in knowledge discovery, not the opposite. They have zero interpretability (good luck figuring out why and what the weights are), which you should know by now had you attended a single Machine Learning class.
You will NOT what caused it to do behavior it shouldn't do.
Claude can't even extract a PDF. Expecting it to write anything but basic for-loops or callbacks is just dumb.
Hell, I have seen ChatGPT hallucinate an Arduino library in two different countries and two different timeframes and students all fell for it.
Sorry to break it to you but outside of HackerNews people are increasingly anti-AI.
But I guess you can peddle AI slop with the hotdogs models, that will show 'em!