points
The goal is none. The current situation: everything that matters requires human intervention.
I think the end situation will be that LLMs will be able to perform decently well in a highly controlled and predictable environment.