Comment by hiAndrewQuinn

No need to set impossible standards here, the correct thing to ask is whether the guardrailed LLM confabulates less than its competitors. That sounds like a useful job to have.