Machine Learning Isn't An End All Be All
Sure machine learning is currently trending at the moment, in light of news about GPT3. However lets break it down.
It seems like so many youtube shows have a very little understanding of artificial intelligence, and use it purely for the purpose of fear mongering. Especially when you actually dig into the code itself, the idea this will ever become ( a non fragile ) ASI is kind of a joke.
Even at portions where it seems like they’re “creating” something new, it comes already built with subroutines it already knows how to do. Now you could download a module off the web I suppose, but that’s really not the same thing as writing your own code.
At best what you might have is the ability to download trained data on your personal cloud ( I use Syncthing rather than Google ) so you don’t have to constantly retrain a machine learning function every time you want to run it.
However machine learning isn’t an end all be all.
There are still functionalities where explicitly programmed code is more optimal than using classifiers that produce labels and probabilities. You might, however need machine learning to do label and probability production.