When you build something from scratch, crafting every component yourself, it can take a long time to finish your product. It would be so much easier to start with ready-made components, from off the shelf, and just snap them together.
Also, given constraints on time and effort, you might get more polished results by taking the ready-made path. But if your goal is to understand how something works, to figure out new and generalizable principles, the “build from scratch” approach has definite advantages.
When you build something yourself, from the ground up, there are no black boxes. You understand how each component works, as well as how and why those components all fit together.
To me the great danger of the trend to use A.I. for everything is precisely this. A.I. will assemble things for you, often from high polished components.
But it won’t give you any understanding or insight. People will end up with shiny artifaces, and on the surface it will all seem great.
But the human advancement of knowledge and insight, the hard work of understanding, will continue to slow. And eventually we will have no idea how anything actually works, nor how to make the intuitive leap required to create something truly new.