Decades ago, long before anyone had heard of a large language model, I wrote programs that responded to a random event (inside a game) like a death of a friend

Decades ago, long before anyone had heard of a large language model, I wrote programs that responded to a random event (inside a game) like a death of a friend by outputting statements that the program itself was grieving. LLMs are doing nothing more advanced than that. There's no justification for trying to blur the lines that make an AI model appear to have emotions.