Because back then you had to care about the differences of e.g. 16bit vs 32bit datatypes & instructions, hardware limitations are not known to modern grown-up dev people - they think hardware is unlimited. Earlier you really had do understand what you were doing, today cou can get along with a prompt - which works great, until it doesnt. Not to blame any vibecode hobbyist here, though real application development usually required some degree of deeper understanding back then.

I think the point is that you had to be deeply curious and more of a "hacker" or "computer nerd" type to be able to figure things out.

But I think the same applies to not just AI but various tools that have abstracted away the complexity of things over the years.

For example, I would imagine the average person deploying some sort of web app or API today knows far less about networking and infrastructure than someone doing it 10 to 20 years ago.

Knowing that genuine, disincentivized creativity is exceedingly rare (especially in the West), you can assume that the answer looks something like a carrot or a stick.

Because it's "easy"?

Because it's "easy" (until they hit a wall)

Once they hit a wall, that is where you find out whether they are motivated or not