It still makes me feel odd when ChatGPT says something like "I feel the same way", "I love xyz", "I think ...", exactly because it can't really feel anything. It defends this behaviour as an abstraction to make the conversation more natural, but that makes it easy to forget that there is no feeling or opinion - only the inputs/outputs resembling one.
It also annoys me that it uses the word "we" when talking about humanity
I think this is the entire point, you're supposed to subconsciously believe that you're talking to a human, not a matrix multiplication algorithm