>That is to say, if a machine understands what you're saying and a human doesn't, that human's criticism might be below engaging with. No wonder OpenAI acciden

>That is to say, if a machine understands what you're saying and a human doesn't, that human's criticism might be below engaging with. No wonder OpenAI accidentally made LLMs even more sycophantic if this is what people want from them.