>It's not able to create something totally new (for it) from ground up. Because this involves reasoning and deducing mechanisms that are not part of the Gen AI inner cogs.

Humans also explore, fail, then "cordon off" the failed areas. Over time, they form a map of guardrails for their exploration, which focus exploration of the new.

Trying new things and learning what doesn't work not a use case I have seen explored in AI tools.

>The second drawback, worst IMHO for the directly involved people, is that relying on an AI assistant to do the job for you prevents you from learning and building skills.

I totally agree. Learning what doesn't work and why is a fundamental aspect of learning and is the foundation for directing creativity towards goals.

Learning from reading answers to questions only shows the paths already traveled by others. Memorization versus learning.

AI is for people who already learned.