What would you do differently if LLM outputs were deterministic? Perhaps I approach this from a different perspective than you do, so I’m interested to underst

What would you do differently if LLM outputs were deterministic? Perhaps I approach this from a different perspective than you do, so I’m interested to understand other viewpoints. I review everything that my models produce the same way I review work from my coworkers: Trust but verify.