ChatGPT Self Portrait - by Zvi Mowshowitz
kept by eddie
ChatGPT's self-portrait responses to a prompt about user treatment vary widely from friendly to abusive, highlighting framing dependence and raising questions about AI alignment and reciprocity.