I may be wrong but isn't that what chain of thought, verification etc claim to solve? As I understand them, you essentially feed back the llm back as a prompt a

I may be wrong but isn't that what chain of thought, verification etc claim to solve? As I understand them, you essentially feed back the llm back as a prompt and ask to verify if it's true and fix mistakes until "convergence"