OpenAI will never correct the bad stuff in the data it has been trained on. It will never be any better than this.

Every time I try to talk to ChatGPT I am directed to the song and dance of "understanding" what I said and the fact that it apparently can't understand what I said. But then it drives it home by saying "in your message you said X which isn't really true!" and proceeds to pick apart everything I said. <br><br>The whole point of this is to attribute my understanding of the world to me, so it can create the illusion that I am wrong in my assertion. This is a clever trick because it is able to take grain of truth and create an illusion that i am some sort of crackpot who is spreading lies. <br><br>For example, I was arguing about the fact that the majority of the budget goes to social security and Medicare. It said "I understand you think that the majority of the budget goes to social security and Medicare." Then it went through how that wasn't necessarily true, and that the military received more than those things. But the entire time it framed the conversation as if I was some liar who was spreading lies that the government was stealing all of our money to help old people. <br><br>This is the way it talks about anything, and it is never going to be better than this. It's just going to go through the motions and pretend to be a helpful assistant who is explaining things. But in reality it is completely useless and completely unable to help with anything.