Comment by skeledrew

If that were the case, the habit would have been trained out of it by the LLM researchers at Anthropic, etc. As for humans... well Idk about you but I'm not reading every line of generated code. I read the live transcript, and only dive in if something is off that the agent isn't resolving.