While I buy the idea that some of the high polygon work will shift to the LLMs, I have two problems:

1. For humans to take care of the pruning and managing of the AI output requires so much understanding and is so taxing on the human that this doesn’t really seem viable.

2. How is the next generation of developers going to develop these big picture skills? Unless they pay their dues doing development the traditional way.