Feedback: I tend to say thank you after an interaction with LLMs. Your model doesn't really understand what I'm asking and gives me more and more troubleshooting data, whereas the expected behavior would be a "you're welcome" or equivalent.
I know its probably not critical path but its one of those things that may help drive people's relationship and mindshare of the platform.
This is really important to me also. ChatGPT is so human-like that I really feel a deep psychological need to say "thank you" to it. I feel awful if I don't. The ironic thing here is that I'm now causing more work by demanding support for my ability to say "thanks" when it's currently not supported. But I say things like, "thanks, this is exactly what I was looking for!" Or "thanks, this was really helpful!" or just, "thanks!" and then I feel better about the encounter. If I couldn't say "thanks" and that confused it I'd be sad so I'd appreciate "thanks" and such variants being understood.
I know its probably not critical path but its one of those things that may help drive people's relationship and mindshare of the platform.