Calling an interface “chat” doesn’t change anything, you still have the same options for communicating information; visual, text, audio, etc. You’re just personifying your UX. Whatever text Christina is “saying” could be included in the UX in different ways. Requiring Christina to “say” it can add undue burden to the text by needing to include various politenesses to maintain the personification illusion, ones that I don’t want to waste my time reading when they come from an AI.
I don’t understand what you mean by “using a conversational style” in the your user testing claim. I would want to see the same wording tested in two conditions; one where the computer has been personified and you have to pretend it’s a person in order to interact with it, and the other where you use conventional text to say the same thing but leave out the personification.
My other complaint about bots is that I don’t want to type in order to interact with them. Click and tapping is much quicker and easier than typing for simple task. When bots can handle the complex tasks I’ll gladly use them if they are indeed easier.
Chat might be an interesting way for designers to start playing with conversational, voiced-based interactions, but we should admit that it is not yet voice-based. Talking to the computer will be great someday. Texting with it, I’m much more skeptical.