One conversation where the user got to make it say weird stuff because he purposely manipulated it does not mean it needs to be taken away from all users. I use it a little bit like a research assistant and it helps tremendously. Do I trust all of its outputs? No, but it gives me a start to look at topics in more detail.