Ai is terrible, specifically Gemini and ChatGPT are bad, they are purposely sycophantic. Gemini over the last week was tweaked to be even worse, it constantly asks for what I think about something at the end of a response. It felt off the first few times so I went and looked at my settings to see if I am feeding the data to train their model. Turns out there is no way to opt out of training, and google will always use our data to train. So they tweaked the responses to get more opinions from the users. Claude is also sycophantic but to a lesser extent.