points
Why do we believe that we cannot train models which could explain the jargon in more human terms when current LLMs can perfectly explain the most complicated codebases?