I agree with you somewhat, or want to, but the problem I have is that there's a vast corpus of knowledge that is only available in "dense books" -- or papers. See for instance most topics in higher math, or virtually any academic subject

YouTube and ChatGPT work when lots of other people have digested that material on your behalf and created educational content. And there will always be people doing that, no doubt, and the amount of content like that will only get better as time goes on, there are lots of excellent YouTubers creating math content about niche topics.

But it reminds me of that instinct of young SWEs meeting a legacy codebase for the first time to just "rewrite it all" -- taking the time to understanding a codebase is a skill, and so is learning to work around imperfection.

Part of learning how to learn is making do with resources that explain things in ways you still find difficult at first brush. Indeed, for entry level math or CS topics theres's often a staggering array of YouTubers/resources to choose from, but for many difficult topics sometimes all you'll have will be a single paper from the 50s where the author is dead -- or a 1000 line SQL script written by an unknown consultant 15 years ago.

So in short -- I think neglecting that skill over a lifetime is going to be harmful for anyone who wants to do novel work or push the boundaries of human knowledge or even most knowledge workers in their day to day. But for most people though, you are probably right that it's just a different approach to finding information