I would just be fascinated by the retro 90s landscape of 3D software and productions.
Typically the general gist of it, would be something this:
• We have this innovative project never done before.
• We have this brand new piece of software never seen before.
• We a 3D person without prior experience.
• We have only a few months to complete the project.
And you get something like this:
[ More or less the gist out of the “how it was made” video. ]
However it goes without saying that everything else was set in place:
• The theme, the lore, the world building, the concept.
• The character designs were curated and perfected.
• Everything got overviewed and reviewed multiple times until perfected.
With this mindset, is very easy to consider that the fundamentals of 3D, are entirely irrelevant to 3D. As for example creating a torus and painting it beige/brown is only a demonstration of technical skill. If you get stuck into creating 20 donuts, or even use geometry nodes, or if you go even further to use gaussian splatting of scanned donuts, you would just not going to make an impact.
Somewhat everybody can fall (or many of us in the past) into this logical fallacy. That learning 3D is the real deal that solves all the problems, but this is only in technical terms.
So in this in this way, I consider that even a beginner can definitely create professional-grade work, only with proper guidance and accurate feedback, after multiple of iterations. So is easy to understand, that is not a matter of “skill” but rather that is a matter of “curating” and “eyeballing”.
I consider that at this time and age (the last 20 years or so), that people have become too technical in their mindset, and this caused huge “monolithic” approach to learning as well as to working. Even if for example some would say that AI fixes the creativity problem, you would still have a problem with “curation” and “nuanced approach”, AI is just speeding up things.
As for example with this type of thinking, if you get a game like Concord that it was autogenerated. I doubt that getting 100^2 additional iterations of it would fix anything. You would just go in circles and wasting time.