The trouble with the "learn just this one thing" approach is that one is forced to learn said thing at the most basic level because anything beyond that requires all sorts of skills and techniques that are difficult to teach in lesson format. Rather, they're just absorbed as one explores the topic. It's the sort of subconscious / muscle memory stuff a person doesn't even realize they're learning.
So, yes, for the most basic of topics, I can see how removing the effort can make sense. For anything beyond that I feel there is tremendous value in the struggle.
Right now you could open a random Wikipedia article, study it and click random again, clearly there’s better options. SO wasn’t quite that bad as it was more constrained, but you still didn’t do it without external pressure to find something.
I’m told some use AI to do boilerplate work, but I bet those that buy wholesale into the AI hype use it for everything, especially things they don’t know much about and would actually benefit from learning themselves.
I am self-taught and naturally curious: having a machine to rob me of the chance to learn and discover, to beat my head against a wall, which I also have to babysit because it is dumb as a rock, is simply a net negative for me.
It reminds me a bit of looking up coding solutions on StackOverflow back in the day. Yes, it was a slog and consumed valuable time. However, I always felt that I picked up all sorts of other valuable information reading a variety of possible solutions or spying a related but different problem and taking a moment to consider it.
It is similar to how I learned to play guitar. With no videos and very limited tablature, I had to learn songs by ear which was crazy inefficient. However, it trained my ear and kept me exploring the entire fretboard as I figured the song out. This ended up making me a much more complete guitarist.