SpaceX is useful because upper management actively subvert him. Tesla had a functional process before he decided to smear his face all over the public image of it. Twitter was one example of interesting architecture before he gutted it (and caused a LOT of problems; I'm informed about 80% of the infra he took out has been put back in).
He's the very real example of an accidental success without any understanding of why he succeeded.
Deleted Comment
This is what is laughable, and has been from the very beginning. Will we _continue_ to use AI to try and find novel drug treatments? Yes. Will chatbots still be the first thing a customer support flow uses (and will it still suck)? Yes.
In my experience Claude and Gemini can take over tool use and all we need to do is tell them the goal. This is huge, we always had to specify the steps to achieve anything on a computer before. Writing a fixed program to deal with dynamic process is hard, while a LLM can adapt on the fly.