I wonder if that makes the code worse on average, considering how LLMs seem to dramatically increase their reasoning capabilities if you ask them to show their

I wonder if that makes the code worse on average, considering how LLMs seem to dramatically increase their reasoning capabilities if you ask them to show their thought process