I tried chatgpt and claude, several times over the past year. To get it to say anything sensical is a laborious game of prompt trial and error, and even then in the end the code is of junior level at best, containing subtle bugs, inefficient instructions and needs a lot of refactoring. Not sure how emotions come into play, I tested a tool and found it useless to me. I'm much faster writing my own code, that I understand deeply, can reason about and learn from. I was never a fan of copy pasting big chunks of stack overflow that I didn't understand, which is exactly what LLM coding tools facilitate.