The solution, then, is to add comments to every difficult line of code and have an LLM check that comments match the code. Then you get precision and reliabilit

The solution, then, is to add comments to every difficult line of code and have an LLM check that comments match the code. Then you get precision and reliability of machine language + readability of human language